Datasets:

CleverShovel

/

paper_reviews

like 1

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: RECOMMENDATION ALGORITHM BASED ON ATTRIBUTED MULTIPLEX HETEROGENEOUS NETWORK Review round: 1 Reviewer: 1

Basic reporting: 1. Novelty of idea is not clear. What is your main impact from the research and results? Make clear presentation also in relation to recent ideas in the field of deep learning. 2. Related models to present: MobileGCN applied to low-dimensional node feature learning, Self-attention negative feedback network for real-time image super-resolution. 3. Revise abstract to better show your main achievement and results. Experimental design: 1. Proposed in fig. 1 – fig. 2 model is not clear. What is the main input to this model? How do you preprocess info on the input? 2. What features are considered by your model? It is not presented how do you make them from the input. Do you use normalization before processing? 3. How is the model of graph in fig. 3 constructed? 4. Why Adam was used in your model? Did you test other algorithms? Validity of the findings: 1. It is not necessary to use probability in eq. (15) – eq. (16). Do you really use probability in your model? Can you show in which place of model it is used? 2. Your model needs comparisons to other model and comparisons on other data. 3. It is not clear what is your research hypothesis for amazon or youtube data. We read about results but what actually do you want to verify? Additional comments: Images need better quality, since resolution is low and will have bad outlook in paper

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: RECOMMENDATION ALGORITHM BASED ON ATTRIBUTED MULTIPLEX HETEROGENEOUS NETWORK Review round: 1 Reviewer: 2

Basic reporting: The paper had proposed the recommendation algorithm based on attributed multiplex heterogeneous network. The attributed multiplex heterogeneous network can help obtain more user-item information with more attributes. The proposed framework SSN_GATNE-T was tested on two different types of datasets, Amazon and YouTube. The proposed algorithm improves the three evaluation indexes, especially the evaluation index F1. I suggest this paper be returned to authors for some related revisions, I hope those follow comments will help the authors to improve this paper. Experimental design: I suggest the authors amend related description and depict with experiments. Validity of the findings: 1. The abstract and conclusion need to be improved. The abstract must be a concise yet comprehensive reflection of what is in your paper. Please modify the abstract according to “motivation, description, results and conclusion” parts. I suggest extending the conclusions section to focus on the results you get, the method you propose, and their significance. 2. What is the motivation of the proposed method? The details of motivation and innovations are important for potential readers and journals. Please add this detailed description in the last paragraph in section I. Please modify the paragraph according to "For this paper, the main contributions are as follows: (1) ......" to Section I. Please give the details of motivations. 3. The description of manuscript is very important for potential reader and other researchers. I encourage the authors to have their manuscript proof-edited by a native English speaker to enhance the level of paper presentation. 4. Please update references with recent paper in CVPR, ICCV, ECCV et al and Elsevier, Springer. In your section 1 and section 2, I suggest the authors amend several related literatures and corresponding references in recent years. For example: The improved image inpainting algorithm via encoder and similarity constraint (The Visual Computer); Research on image inpainting algorithm of improved total variation minimization method (Journal of Ambient Intelligence and Humanized Computing); The image annotation algorithm using convolutional features from intermediate layer of deep learning (Multimedia Tools and Applications); Image super-resolution reconstruction based on feature map attention mechanism (Applied Intelligence). 5. Please check all parameters in the manuscript and amend some related description of primary parameters. In section 3, please write the proposed algorithm in a proper algorithm/pseudocode format with Algorithm. Otherwise, it is very hard to follow. Some examples here: https://tex.stackexchange.com/questions/204592/how-to-format-a-pseudocode-algorithm 6. The section 2 is too short, and I suggest the authors amend the details of background and motivation. The main section of manuscript is section 3. I suggest the authors amend related depict of proposed method. Additional comments: no

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: RECOMMENDATION ALGORITHM BASED ON ATTRIBUTED MULTIPLEX HETEROGENEOUS NETWORK Review round: 1 Reviewer: 3

Basic reporting: Recommender systems have been widely used in the real world. In addition to collaborative filtering, heterogeneous information networks (HINs) have attracted more attention in recent years. HIN-based recommendation algorithms have also been proved to be effective in a few practical recommendation scenarios. This work combines HINs and multiplex networks, a compound network made up of two or more sub-networks, to construct an attributed multiplex HIN for item recommendations. Besides, this work proposes an attention mechanism that can work well with the proposed network to recommend appropriate items for the target user. Basic reporting comments include: 1. This article is well organized. The presentation of this paper can be further improved with the help of native speakers of English. 2. All the figures of this article could be replaced with high-resolution ones. 3. In the Introduction section, the motivation of this work is not clear. Why did the work combine HINs and multiplex networks? The authors should point out the background, shortcomings of the existing models, and challenges. 4. A few key references of the related work were not included and analyzed in the second section. The authors should carefully review the studies on HIN-based embedding and multiplex network embedding and add those missing works in the revised manuscript. [1] Fenfang Xie, Angyu Zheng, Liang Chen, Zibin Zheng: Attentive Meta-graph Embedding for item Recommendation in heterogeneous information networks. Knowl. Based Syst. 211: 106524 (2021) [2] Léo Pio-Lopez, Alberto Valdeolivas, Laurent Tichit, Élisabeth Remy, Anaïs Baudot: MultiVERSE: a multiplex and multiplex-heterogeneous network embedding approach. Sci. Rep. 11: 8794 (2021) [3] Qiwei Zhong, Yang Liu, Xiang Ao, Binbin Hu, Jinghua Feng, Jiayu Tang, Qing He: Financial Defaulter Detection on Online Credit Payment via Multi-view Attributed Heterogeneous Information Network. WWW 2020: 785-795 [4] Zhijun Liu, Chao Huang, Yanwei Yu, Baode Fan, Junyu Dong: Fast Attributed Multiplex Heterogeneous Network Embedding. CIKM 2020: 995-1004 [5] Yukuo Cen, Xu Zou, Jianwei Zhang, Hongxia Yang, Jingren Zhou, Jie Tang: Representation Learning for Attributed Multiplex Heterogeneous Network. KDD 2019: 1358-1368 [6] Binbin Hu, Zhiqiang Zhang, Chuan Shi, Jun Zhou, Xiaolong Li, Yuan Qi: Cash-Out User Detection Based on Attributed Heterogeneous Information Network with a Hierarchical Attention Mechanism. AAAI 2019: 946-953 [7] Yiming Zhang, Yujie Fan, Yanfang Ye, Liang Zhao, Chuan Shi: Key Player Identification in Underground Forums over Attributed Heterogeneous Information Network Embedding Framework. CIKM 2019: 549-558 5. The concept of multiplex networks has been formally defined in the field of network science. However, the authors did not formulate it in this article. How is the attention mechanism applied to the proposed network that is essentially a multiplex network? Experimental design: This study experimented on two real-world datasets and demonstrated the effectiveness of the proposed approach in terms of three evaluation metrics. The reviewer’s concerns on experimental design include: 1. The selected baselines have been proposed for three years. The authors should add more recent approaches (e.g., the above refs. [2], [4], and [5]) based on attributed multiplex heterogeneous networks to compare with the proposed approach. 2. In addition to the parameter setting of the proposed approach, the authors should introduce the settings of each baseline’s hyperparameters, which may impact the performance of each baseline. 3. The implementation code of the proposed approach is not available for research on the Internet. 4. In addition to the ablation experiment, the authors should discuss the impacts of some critical hyperparameters (e.g., the dimension of embeddings and the layer number of a multiplex network) on model performance. Validity of the findings: 1. The motivation of the Fridman test is not clear. For example, what are the actual meanings of node1 and node2? Why did the Fridman test on the two datasets obtain two different results, i.e., the original hypothesis was retained on the Youtube dataset, while the original hypothesis was rejected on the Amazon dataset. 2. If possible, the authors should add a subsection to discuss some potential threats to the validity of this work, including construct validity, internal validity, and external validity. Additional comments: No comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: RECOMMENDATION ALGORITHM BASED ON ATTRIBUTED MULTIPLEX HETEROGENEOUS NETWORK Review round: 2 Reviewer: 1

Basic reporting: i think paper is revised Experimental design: i think paper is revised Validity of the findings: i think paper is revised Additional comments: i think paper is revised

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: CONTEXT-BASED SENTIMENT ANALYSIS ON CUSTOMER REVIEWS USING MACHINE LEARNING LINEAR MODELS Review round: 1 Reviewer: 1

Basic reporting: Clear and unambiguous Literature review is fair Figures and Tables are clear Experimental design: Novelty is lacking only application of existing algorithm with GPU based infrastructure Validity of the findings: Good choice of algorithm and datasets Additional comments: 1. Paper does not reveal whether multiple modalities have been used for sentiment Analysis. It can be mentioned. 2. Sarcasm is not part analysis and can be added

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: CONTEXT-BASED SENTIMENT ANALYSIS ON CUSTOMER REVIEWS USING MACHINE LEARNING LINEAR MODELS Review round: 1 Reviewer: 2

Basic reporting: 1. The article is clear in understanding and standard structure has been used. 2. Sufficient literature has been surveyed but lacks standard journal references like IEEE transactions, Elsevier springer etc. 3. The data set used is standard datasets, Experimental design: 1.Sentiment analysis using machine learning is one of the upcoming area and is within aim and scope. 2. Extensive investigation is performed with machine learning and deep learning models. 3. I could see only fasttext comparison with amazon review and all other datasets mentioned, but LSVM and SA-BLSTM of existing literature is not shown for amazon, yelp dataset. Validity of the findings: 1. findings provided in the table are valid. 2. Conclusion should highlight the accuracy measure of the proposed model Additional comments: Highlight what is the % of performance evaluation in abstract also

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: CONTEXT-BASED SENTIMENT ANALYSIS ON CUSTOMER REVIEWS USING MACHINE LEARNING LINEAR MODELS Review round: 2 Reviewer: 1

Basic reporting: 1. The article is clear in understanding and standard structure has been used. 2. Sufficient literature has been surveyed but lacks standard journal references like IEEE transactions, Elsevier springer etc. 3. The data set used is standard datasets, Experimental design: 1.Sentiment analysis using machine learning is one of the upcoming area and is within aim and scope. 2. Extensive investigation is performed with machine learning and deep learning models. Validity of the findings: 1. findings provided in the table are valid. 2. Deep learning method has been tried. Additional comments: 1. All suggestions mentioned in the previous review has been incorporated.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SIM-DRS: A SIMILARITY-BASED DYNAMIC RESOURCE SCHEDULING ALGORITHM FOR MICROSERVICE-BASED WEB SYSTEMS Review round: 1 Reviewer: 1

Basic reporting: This paper tackles the resource scheduling problem in web systems using a dynamic resource allocation/scheduling method, which has long been a prevalent model and commonly adopted for microservice composition in the domain of distributed computing. The paper formulated a combinatorial optimization problem and proposed a heuristic scheduling algorithm to find feasible scheduling schemes by directly adopting existing solutions to similar problems. Experimental design: A benchmark named TeaStore is used for experimental evaluation. TeaStore has been widely used for performance evaluation of microservice-based applications. Validity of the findings: The performance evaluation shows that the proposed approach outperforms other methods, which is quite expected since the existing methods (which are known to be better) are used. Additional comments: Resource scheduling in distributed computing environment is well-known to be of interest of researchers in the related area and has been investigated extensively in literatures. Although this paper includes some interesting ideas and results, the following issues need to be addressed before acceptance. Strengths: 1. This work offers an interesting idea, i.e., solutions to similar problems often share certain structures, which is instructive for solving scheduling problems and could avoid unnecessary overhead of solving problems from scratch. 2. The performance evaluation shows positive results that make the idea worthy of further investigation and adoption. Weaknesses: 1. A formal complexity study of the formulated problem “DRS-MWS” is missing. 2. The related work section should be better summarized, e.g., the “rule-based approach” and “heuristic approach” contains too much work in a single paragraph and these two approaches may contain some overlap and their distinction is not well described. Also, a complete refresh of the related work with more recent literatures would be appreciated. 3. Eq. 9 seems to be messed up, please correct

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SIM-DRS: A SIMILARITY-BASED DYNAMIC RESOURCE SCHEDULING ALGORITHM FOR MICROSERVICE-BASED WEB SYSTEMS Review round: 1 Reviewer: 2

Basic reporting: no comment Experimental design: no comment Validity of the findings: no comment Additional comments: This paper presents Sim-DRS, a similarity-based heuristic scheduling algorithm for microservice-based web systems that aims to find viable scheduling schemes by utilizing solutions to similar problems. More specifically, the Dynamic Resource Scheduling for Microservice-based Web Systems (DRS-MWS) is formulated as a combinatorial optimization problem, and the authors propose the Sim-DRS to solve the DRS-MWS problem. The experimental results show the superiority of Sim-DRS over three state-of-the-art scheduling algorithms. Overall, the article is well organized and its presentation is good. The interesting idea is to obviate a fresh start in dynamic resource scheduling by incorporating previously-known viable solutions to similar problem. However, some minor issues still need to be improved: 1. The introduction section should firstly provide an idea of the resource scheduling process before the analysis of existing methods. 2. The words in Figure 1 are not very clear, and the contents in Table 2 and Table 3 are not clearly displayed. Please improve the quality of figures and tables. 3. There are some typos and grammar errors. It is recommended that the authors should proof read the manuscript before submission.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SIM-DRS: A SIMILARITY-BASED DYNAMIC RESOURCE SCHEDULING ALGORITHM FOR MICROSERVICE-BASED WEB SYSTEMS Review round: 1 Reviewer: 3

Basic reporting: The paper is easy to read and the whole process is complete and neat. More specifically, the contribution of the work consists of three key components: (1) This paper describes DRS-MWS as a combinatorial optimization problem, which clarifies the processing logic of the original problem; (2) This paper proposes Sim-DRS to solve DRS-MWS. It uses prior knowledge, that is, to directly use feasible solutions to similar problems to find a promising scheduling plan, thus effectively improving the efficiency of the algorithm; (3) This article uses a well-known microservice benchmark called TeaStore to evaluate the performance of Sim-DRS through extensive experiments under different scheduling time constraints. Experimental design: The Sim-DRS works well in experiments, and the research results obtained should be very effective, very useful to practitioners and to promote future research directions.. Validity of the findings: This paper proposes Sim-DRS, which uses the identification and combination of previously known feasible solutions to similar problems as initial values, and uses evolutionary algorithms to quickly find promising decision-making solutions for dynamic resource scheduling. Additional comments: To improve the paper, I have some comments though: 1. There are some formatting issues in the paper. For example, the in formula (6) should be left aligned; the result ( ) in formula 9 can try to give the reader. 2. Some symbols are not clearly defined. For example, the " " in equation (8) is unclear. If it is the dot product of the matrix, it is recommended to use " ", if it is the outer product or Hadamard product, it is also recommended to use the corresponding mainstream symbols to express; it is recommended to consider replacing (in line 180) with for more clarity. Similar issues are also recommended to be modified, etc. 3. Unify the expression of images in the text, such as "Figure 3" in line 233 and "Fig. 5" in line 330.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SIM-DRS: A SIMILARITY-BASED DYNAMIC RESOURCE SCHEDULING ALGORITHM FOR MICROSERVICE-BASED WEB SYSTEMS Review round: 2 Reviewer: 1

Basic reporting: My previous comments have been addressed in this version. I do not have any further comments. Experimental design: My previous comments have been addressed in this version. I do not have any further comments. Validity of the findings: My previous comments have been addressed in this version. I do not have any further comments. Additional comments: My previous comments have been addressed in this version. I do not have any further comments.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SIM-DRS: A SIMILARITY-BASED DYNAMIC RESOURCE SCHEDULING ALGORITHM FOR MICROSERVICE-BASED WEB SYSTEMS Review round: 2 Reviewer: 2

Basic reporting: No Experimental design: No Validity of the findings: No Additional comments: No

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: COMPARATIVE ANALYSIS OF MACHINE LEARNING APPROACHES TO ANALYZE AND PREDICT THE COVID-19 OUTBREAK Review round: 1 Reviewer: 1

Basic reporting: The current study does not explain why these models were selected for comparison. Specifically, it lacks a review or summary of the existing studies that have applied forecasting models for the COVID-19 pandemic. Please see the existing papers in PubMed/LitCovid (https://www.ncbi.nlm.nih.gov/research/coronavirus/docsum?text=forecasting%20model). The methods are from traditional mathematical models and recurrent neural networks other than ANN. The study should discuss them in detail and select representative models with reasonings. Experimental design: First, the study should explain the data split and cross-validation in more detail. How was the cross-validation performed? Since this is time-series data, the split by time is critical. How does the 10-fold cross-validation reflect the timeline? Second, a more religious design should be made for the forecasting model. Currently, the spanning between the train and testing is too close. The training set contains the data up to 7th Nov 2020, whereas the testing data is directly from 8th Nov 2020. Arguably, if the data of 7th Nov 2020 is available, it is not too challenging to predict the 8th even using a simple model. And it is not very useful because there is not much time for interventions or prevention can be done in advance. The study could quantify how the model forecasts in a week or earlier in advance. Validity of the findings: The study lacks a thorough discussion on the limitations. For example, there are many factors (prevention, vaccinations etc) that could impact the case growth. Such factors cannot be captured in the models. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: COMPARATIVE ANALYSIS OF MACHINE LEARNING APPROACHES TO ANALYZE AND PREDICT THE COVID-19 OUTBREAK Review round: 1 Reviewer: 2

Basic reporting: The manuscript size should be reduced, by improving the writing style. For example, information included in Tables, should not be also mentioned in the main body of the manuscript in such detail. E.g., lines 269-281, lines 283-296, lines 301-313, etc. Replace these details with a more qualitative and descriptive discussion of the overall findings. The literature review on the topic is very poor at the moment, and should be enhanced. The connection of this work with previously published work in the Journal should be established. The manuscript needs very thorough editing and proofreading. It would be best to remove the gray background of the figures. The green in Figure 2 is too bright; replace it with a smoother one (use the same blue and red as in Figure 1). Experimental design: The research question is well defined, and it contributes to the international literature on the topic. The methods are adequately described. Validity of the findings: This work has merit to be an important contribution to the topic of using machine learning methods for COVID-19 forecasting, by comparing several such methods. Additional comments: This is an interesting manuscript that has a contribution to the international literature of COVID-19 forecasting, by comparing machine learning approaches towards this direction. There are some issues that need to be addressed before the manuscript is ready for publication.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: COMPARATIVE ANALYSIS OF MACHINE LEARNING APPROACHES TO ANALYZE AND PREDICT THE COVID-19 OUTBREAK Review round: 2 Reviewer: 1

Basic reporting: The reviewers have addressed my concerns. Experimental design: Thanks, reviewers for the efforts in addressing my comment #3. However, this comment has not been fully addressed. The statement 'Actually dividing the data into training and testing parts is used to evaluate the model performance. The model that are trained and the data are tested is completely unseen to the model. ' made by the author is fine but is not enough for the evaluation on generalization. Splitting training and testing from the same dataset of course is a way to evaluate the performance. However, splitting from the same dataset will make the training and testing set share similar distributions or characteristics, which is not always the case when applying to new data. In this specific case of outbreak prediction, it is critical to apply to an external dataset that does not have overlapped periods to the training and testing datasets. Validity of the findings: I do not have additional comments. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: COMPARATIVE ANALYSIS OF MACHINE LEARNING APPROACHES TO ANALYZE AND PREDICT THE COVID-19 OUTBREAK Review round: 3 Reviewer: 1

Basic reporting: My last concern has primarily been addressed. Thanks authors for their dedicated efforts Experimental design: N/A Validity of the findings: N/A Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 1 Reviewer: 1

Basic reporting: - Please improve the overall readability of the paper. - The objectives of this paper need to be polished. Contribution list should be polished at the end of the introduction section and last paragraph of the introduction should be the organization of the paper. - In the first four paragraphs of literature review section, the authors have presented a good references, but they need to present the recent and most updated references. - Some Paragraphs in the paper can be merged and some long paragraphs can be split into two. (i.e., Related Work first paragprah of more than 2pages) - Remomve the authors name from table 1 first column. Just give reference number. Experimental design: - Please cite each equation and clearly explain its terms. - Clearly highlight the terms used in the algorithm and explain them in the text. - The procedures and analysis of the data are seen to be unclear. Validity of the findings: - What are the evaluations used for the verification of results? - Provide a comparison with existing studies. Additional comments: - Make sure the Conclusion succinctly summarizes the paper. It should not repeat phrases from the Introduction! - Authors should add the most recent reference: 1) CANintelliIDS: Detecting In-Vehicle Intrusion Attacks on a Controller Area Network using CNN and Attention-based GRU, IEEE Transactions on Network Science and Engineering 2) Anomaly Detection in Automated Vehicles Using Multistage Attention-Based Convolutional Neural Network, IEEE Transactions on Intelligent Transportation Systems lastly, Proofread the paper

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 1 Reviewer: 2

Basic reporting: - Reference format must be standard in the whole paper. Experimental design: - It is recommended that more information adds for simulation environments, parameters and data. Validity of the findings: It is ok. Additional comments: - "VANET security section" needs more detail and references for challenges and criteria identification. - Response to reviewer must be in an individual file.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 1 Reviewer: 3

Basic reporting: propose an online/offline lightweight authentication scheme for VANET cross-domain system in IIoT. The proposed scheme utilizes an efficient AES-RSA algorithm to achieve integrity and confidentiality of the message. The offline joining is added to avoid remote network intrusions and the risk of network service interruptions. The Burrows Abdi Needham (BAN logic) logic is used to prove that this scheme is mutually authenticated. The system’s security has been tested using the well-known AVISPA tool to formally evaluate and verify its security. Experimental design: propose an online/offline lightweight authentication scheme for VANET cross-domain system in IIoT. The proposed scheme utilizes an efficient AES-RSA algorithm to achieve integrity and confidentiality of the message. The offline joining is added to avoid remote network intrusions and the risk of network service interruptions. The Burrows Abdi Needham (BAN logic) logic is used to prove that this scheme is mutually authenticated. The system’s security has been tested using the well-known AVISPA tool to formally evaluate and verify its security. Validity of the findings: propose an online/offline lightweight authentication scheme for VANET cross-domain system in IIoT. The proposed scheme utilizes an efficient AES-RSA algorithm to achieve integrity and confidentiality of the message. The offline joining is added to avoid remote network intrusions and the risk of network service interruptions. The Burrows Abdi Needham (BAN logic) logic is used to prove that this scheme is mutually authenticated. The system’s security has been tested using the well-known AVISPA tool to formally evaluate and verify its security. Additional comments: • The authors should emphasize the difference between other methods to clarify the position of this work further. • The Wide ranges of applications need to be addressed in Introductions • The objective of the research should be clearly defined in the last paragraph of the introduction section. • Add the advantages of the proposed system in one quoted line for justifying the proposed approach in the Introduction section. • The motivation for the present research would be clearer, by providing a more direct link between the importance of choosing your own method. Under blockchain background, the following papers can be referred A survey on blockchain for big data: Approaches, opportunities, and future directions Toward Blockchain for Edge-of-Things: A New Paradigm, Opportunities, and Future Directions

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 1 Reviewer: 4

Basic reporting: 1.Suggestion: Figure 1- The typical architecture of VANETs in page 2; TA has registered OBUs and RSUs. This can be added to the picture. 2. Suggestion: Page 3; Line#105,106 In the offline authentication, TA is not involved in the joining procedure since the information is preloaded prior. The message is not coming out clearly what is been preloaded in advance. A pictorial representation can help 3. Correction: Page 9, Figure 3 - The AES-RSA algorithm work diagram; The final outcome after the comparison of private key and public key comparison is Data is secure not Date is secure 4. Correction: Page 13, The DAT then concatenated the signature with the message CTDTA−→RSU = Enc DTAaes {Signdta ∥ M′ }; The DTA (Domain Trust Authority) not DAT 5. Grammar: Page 16, the movement from RSU1 to RSU2 is occur dynamically; can be changed to does occur or is occurring Experimental design: Question: Page 26; Line 726 - 730 does every RSU preload with vehicle credentials (rvj,SKvj,texp,Ri,TIDv) . What if there are 1000s of RSUs and 1000s of OBU crossing every day? Recommendation: Please quote an example for man in the middle attack or traceability is a concern with an existing VANETs Validity of the findings: no comment Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 2 Reviewer: 1

Basic reporting: - Experimental design: - Validity of the findings: - Additional comments: Accept

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 2 Reviewer: 2

Basic reporting: - It is recommended to check the formats of figures, tables, and references in the journal's standard format for more readability. - It is recommended that short sentences be written to make the article more readable. - In the abstract and introduction, attributes are used relatively, for example, "more security" or "less costly." To better understand, it is necessary to either mention the studied parameters or compare them numerically and as a percentage. - In the previous works section, it is necessary to check the table with explanations. Some papers are not mentioned in the table. For example, in line 178, a source from 2021 is given, which is not mentioned in the table. Experimental design: no comment Validity of the findings: - For equations used in section 6.1, no reference is mentioned, and on the other hand, they are used to compare methods. Therefore, if the authors extract the equations, they add other standard parameters for evaluations. Additional comments: Given that there is no reference to equations, it seems that the evaluation part needs to be improved

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 2 Reviewer: 3

Basic reporting: The authors have addressed all of my comments, paper can be accepted in the current form. Thank you for giving me this opportunity. Experimental design: Good Validity of the findings: Good Additional comments: The authors have addressed all of my comments, paper can be accepted in the current form. Thank you for giving me this opportunity.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 2 Reviewer: 4

Basic reporting: Suggested contents have been modified. Looks good Experimental design: Suggested contents have been modified. Looks good Validity of the findings: data looks good. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A LIGHTWEIGHT AND SECURE ONLINE/OFFLINE CROSS-DOMAIN AUTHENTICATION SCHEME FOR VANET SYSTEMS IN INDUSTRIAL IOT Review round: 3 Reviewer: 1

Basic reporting: No comment Experimental design: No comment Validity of the findings: No comment Additional comments: Authors have addressed the reviewer’s comments satisfactorily.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A 3D DEEP LEARNING APPROACH TO EPICARDIAL FAT SEGMENTATION IN NON-CONTRAST AND POST-CONTRAST CARDIAC CT IMAGES Review round: 1 Reviewer: 1

Basic reporting: This research suggests a U-net based CNN segmentation for epicardial fat, which is very challenging task. Overall, the submitted manuscript seems to have reasonably described the related works and the necessity of the research. However, I have a few concerns about the methods, the evaluation , and the discussion . First, the materials and methods part does not seem to be written clearly. The author had better describe more clearly the contribution of their research. And, I suggest the authors that the materials and Methods part should have some sub sections for readability. Second, I am not sure why the author presented the experimental result of 5-fold cross validation. Are the other experimental results not the cross validation results? Then, the author had better report how to divide the training set and the test set. Currently, it does not seem to be clear whether all the experimental results were cross validation based results or not. Third, in the discussion part, the author had better analyze and infer why such promising results were produced. For example, attention gates and deep supervision module were used in your research. The author had better describe why those gate and module made the performance improved precisely. Experimental design: As I mentioned above, the author presented the experimental result of 5-fold cross validation. Are the other experimental results not the cross validation results? The final result reported in Table 3 are different from the results of Table 2. Are they separate ones? If they are separate ones, I wonder how the author choose the optimal weight in experiments of Table 2. Validity of the findings: The research presented very promising segmentation results. However, it should be also noted that the number of the training case is not large enough. So, I think this should be clearly reported in the manuscript. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A 3D DEEP LEARNING APPROACH TO EPICARDIAL FAT SEGMENTATION IN NON-CONTRAST AND POST-CONTRAST CARDIAC CT IMAGES Review round: 1 Reviewer: 2

Basic reporting: - In summary, this paper presents a computational strategy for automatic segmentation of Epicardial fat (ECF) volume using deep learning computer vision, where the method employs non-contrast and contrast-enhanced CT images. It is a well-organized and well-written paper. - Suggestion and Comments: 1- I would encourage the authors to clearly list the "Clinical Significance" and "Technical significance" of their work at somewhere within the "Introduction". 2- 3D UNet and even 3D attention-based UNet have been around for several years in the community. What kind of novelties this papers is carrying? 3- While for 2D images the UNet network(s) are computationally efficient, we know that 3D convolutions have huge storage requirements and therefore, end-to-end training is limited by GPU memory and data size. Have authors tried 2.5 UNet? If not, why? 4- The current work reports that it employed 15 radiologist to manually segment the data. However, I was not able to find the analysis of inter-rater agreement! I mean, the level of agreement among annotators. I would very much like to see the IoU and/or Dice coefficient to better understand the level of agreement. 5- How the authors did split the 220 patients to build a training, testing, and validation set? 6- I would encourage the authors to provide another comparative analysis using different number of slices plus different image scaling. Experimental design: - Inter-rater agreement among those 15 radiologist is missing. - Data stratification is missing. - In general, the amount of data seems to be few, particularly to tackle one of the most difficult medical image segmentation tasks. Have they tried any type of data augmentation, at least for the training set? - Is the improvement level statistically significance? Validity of the findings: To me, this problem is one the most challenging problem in medical image analysis and interpretation. To me, I expect to see more in deep implementation, more data that covers more diversity, and a generalize model that can be applied across different healthcare centers. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A 3D DEEP LEARNING APPROACH TO EPICARDIAL FAT SEGMENTATION IN NON-CONTRAST AND POST-CONTRAST CARDIAC CT IMAGES Review round: 2 Reviewer: 1

Basic reporting: The authors sincerely responded the reviewers' comments and modified the manuscript. Thank you for the opportunity to review this manuscript. Experimental design: I wish the authors could include the analysis of inter-rater agreement. Except that, I have no comment. Validity of the findings: no comment Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A 3D DEEP LEARNING APPROACH TO EPICARDIAL FAT SEGMENTATION IN NON-CONTRAST AND POST-CONTRAST CARDIAC CT IMAGES Review round: 2 Reviewer: 2

Basic reporting: The authors revised their manuscript accordingly based on my comments and recommendations. Experimental design: The authors revised their manuscript accordingly based on my comments and recommendations. Validity of the findings: The authors revised their manuscript accordingly based on my comments and recommendations. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: ONE FOR “ALL”: A UNIFIED MODEL FOR FINE-GRAINED SENTIMENT ANALYSIS UNDER THREE TASKS Review round: 1 Reviewer: 1

Basic reporting: - The writing of he paper can be improved across the entire paper. I am providing suggestions and line numbers for the authors to make some of the corrections. - The main issue of the paper which made it very hard for me to read is that it is very wordy and the tables and the figures are only appearing at the end. So it's very hard to read the text on results without looking at numbers and figures. I strongly recommend that authors address this. - Authors can also break text into paragraph in places where a paragraph is too long and provide a title for the paragraph to separate different aspects or the parts of the discussion. This is specifically needed when reporting results. - I find the case study section very difficult to read and it also focuses on one or two examples. It does not really provide much insight. Also, you have only included examples where the PEA model does well. It is much more helpful to know which examples it can not get right. First of all, if you bring the relevant table to be next to the section, it can make it easier to understand the text. But in general, if you want to gain extra space, I would remove this section. - When reading through the paper, it's not clear from the method section that PEA will be a unified approach for all the tasks. Bringing Figure 1 to the beginning of the paper will help clarify this. - There are some exceptions for different tasks and datasets but these are only mentioned when reporting the results. These should be in the experimental set up at least. Detailed Comments: - in all the paper ‘pre-trained language representation model’ can be replaced with ‘pre-trained language model’ - Line 115: it’s strange “which makes both advantages of RNN-based models and BERT-based models with ensemble methods” - Lines: 128-131 can be rephrased: graph neural networks (ref1, ref2) have been applied to the problem. - Also, stacked LSTM can be mentioned closer to the LSTM related work. - Related work: the term ‘associates’ has been used a lot, it’s best to rephrase some sentences to avoid this. - If the formatting allows, it’s a good idea to have section or paragraph titles in the related work to separate different aspects. - In the methods section, PEA is the name of the entire solution including data augmentation. It’s better to do a separation: Data Augmentation, Baseline Models, etc. - In line 215: the fusion strategy should not be part of the basic models and part of the proposed model. - 231: we invented is a strange word, maybe use propose. - Equation (1), |entity_i| is a misleading notation. Maybe we can say: |mention(entity_i)| - 247: Table 2 shows an example … - Lines 260-262, the sentence starting with ‘However’, I don’t understand what we mean by it. Can you please rephrase - 272: ‘T is the length of the post, which is the number of words in the post’, the two are redundant, use one. - 290: we conduct experimental attempts to determine the settings - 296: basic models: basic can be dropped - 307: ‘Detailed explanations of CEA can refer to the original paper’: for detailed explanation of CEA, refer to the original paper - 311: Here, you mention that you deal with all the three tasks using the same architecture. You need to make it more clear early on. - 317: It achieved good results in .. -> I has achieved good results (you should add references on tasks where BERT has achieved good results) - 347: sentimental should be just sentiment -390: ‘experimental attempts’ -> experiments - 401: ‘researches’ -> research (replace in all the paper) - 431: ‘while PEA can have stable performance’ what do we mean by stable performance? - 417: you need to have a new section called Results. All the paragraph names should have ‘Results’ instead of ‘Evaluation’ - 400-416: In my opinion, the definition of the metrics (equations) can be removed. But do explain what macro F1 is (F1 averaged over all the classes). - 447: This result shows that the effectiveness -> This result shows the effectiveness - 450: ‘This may be due to the prediction of PEA comes from’-> This may be because the prediction ….. - 452: what is the point of (3), it’s not clear. Do you mean that the increase in performance is not very high? - 454: ‘aspect noise injection is removed for this experiment’ : this is the first time this is mentioned. It should have been mentioned in the experimental set up. - 488: the definition of p-value can be improved. What does ‘The -value is set as 0.05’ means, I think that can be removed. - Lines 495-499: very confusing as to which ones are significant and which one is not significant. Which models does this sentence refer to ‘The estimated -value between these two models is 0.0174, which is lower than 0.05’? - 502-503: ‘Experimental results in section “Experiments and Analysis” illustrated the PEA model’s outstanding performance in ABSA, TABSA and MEABSA than the baselines.’ -> experimental results so far show that the PEA approach is superior to the baselines on all the ABSA, TABSA and MEABSA on selected datasets. - 559: ‘It is introduced that there are two challenges’ -> ‘There are two challenges’ - 575: instead of ‘seriously’, use another word, maybe sharply or massively. - 576: ‘​​This shows that combine the’ -> This shows that the combination of .. - 611: ‘Moreover, we first-time discovered and defined’ -> For the first time, we analysed the effect of the sentiment polarity bias’ - Table 4 is not necessary, you can remove it. Experimental design: Experimental design seems fine. One comment is that how did the authors came up with the weights 0.5 for the ensemble model? Is it possible to optimize this hyper-parameter? Validity of the findings: The findings seem valid and they will be easier to understand if the reporting improves. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: ONE FOR “ALL”: A UNIFIED MODEL FOR FINE-GRAINED SENTIMENT ANALYSIS UNDER THREE TASKS Review round: 1 Reviewer: 2

Basic reporting: no comment Experimental design: Entity replacement is good for low-resource problem, it's also good for the polarity bias problem. It would be better to give some examples of entity replacement results on polarity biased entities. Validity of the findings: I'm interested in the application of this model. If we want to apply the model to real scenarios, what should the data be like to input to the model. Additional comments: 1. The low-resource and polarity bias problems are indeed two challenges in the research field of fine-grained sentiment analysis. The experiments show good results. 2. The design of "dual noise injection" is interesting, it models even non-existed entities and aspects. This improves the generalization ability of the model and the experiments verified it. 3. The case study setting is good for understanding the strength and weakness of RNN-based model and BERT-based model, and explains why the two models should be combined.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: ONE FOR “ALL”: A UNIFIED MODEL FOR FINE-GRAINED SENTIMENT ANALYSIS UNDER THREE TASKS Review round: 1 Reviewer: 3

Basic reporting: 1. Expansion of abbreviations like CEA, RNN, TD-LSTM, ATAE-LSTM should be given the first time the abbreviation is used 2. There are several works on Aspect-Based Sentiment Analysis, but very few on Targeted Aspect-Based Sentiment Analysis and Multi-Entity Aspect-Based Sentiment Analysis. So, literature review on TABSA and MEABSA could be elaborated in detail. Even ABSA literature work could be detailed 3. Racial and gender bias is discussed in literature review, is it implemented in the proposed work? Experimental design: 1. The various attributes of the datasets can be explained, so as to keep the reader engaged. Validity of the findings: 1. According to this work, what is the scope of fine – grained sentiment analysis? How do you differentiate it with coarsely grained sentiment analysis? Explain it beforehand 2. Figure is cited in text, which makes few things unexplained. For example, Figure 1 gives a prediction as output, What is predicted here? Additional comments: 1. Few more examples can be added to explain ABSA, TABSA and MEABSA for better clarity.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: ONE FOR “ALL”: A UNIFIED MODEL FOR FINE-GRAINED SENTIMENT ANALYSIS UNDER THREE TASKS Review round: 2 Reviewer: 1

Basic reporting: no comment Experimental design: no comment Validity of the findings: no comment Additional comments: no comment

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EXPERIMENTAL INTERPRETATION OF ADEQUATE WEIGHT-METRIC COMBINATION FOR DYNAMIC USER-BASED COLLABORATIVE FILTERING Review round: 1 Reviewer: 1

Basic reporting: The manuscript contains a lot of grammatical inadequacies which makes it practically impossible for me to logically follow its flow. I will recommend that the authors should seek the assistance of a native English speaker to help them in editing the manuscript as it cannot be publish in its present form. The entire manuscript should be written in good English language and re-submitted for review. Experimental design: Manuscript has to be re-written first in good English language before comments can be made in this section. Validity of the findings: Manuscript has to be re-written first in good English before comments can be made in this section. Additional comments: Manuscript has to be re-written first in good English language before comments can be made in this section.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EXPERIMENTAL INTERPRETATION OF ADEQUATE WEIGHT-METRIC COMBINATION FOR DYNAMIC USER-BASED COLLABORATIVE FILTERING Review round: 1 Reviewer: 2

Basic reporting: The language needs to be improved. A non-exhaustive list of stylistic glitches and suggestions are: -- Abstract - 24 "recommendations that.." --> "finding recommendations which appeal to each user varies" - 26 "it is measured" --> "we measure the appropriateness of the recommendation in terms of" -- Introduction - 58 "there are loads of" --> "there are many reported RS implementations" - 58 "is blurry" --> "it is unclear, how the " --- Getting started to experiments --> "Experimental design" Other issues similar to the above. The figures Fig 1 to 5 are illegible and it is impossible to verify the conclusions drawn from them. Experimental design: The authors have championed the cause of leave one out methods, however, there should be a discussion of standard statistical augmented methods like ANOVA; that is, the approach discussed here is only true under the severe assumption that each covariate is independent of the others. It is unclear how the median is considered to reduce the outliers; standard techniques to identify (like checking residuals and standardized residuals) should be compared. The median values are indeed better than the mean under certain scenarios, however, it is best to recall, that there are no general unbiased estimators for determining the population median. Note that the Pearson correlation is based on the mean and so the estimator is actually an unbiased estimate of the population statistics. Validity of the findings: It is unclear why the smooth coloring of the JAC with SW is of merit. In general, the distribution of metrics is unclear as a statistic of interest, as it is sensitive to the order of the table and the dataset. If the question is of tracking the metrics themselves, then it would be better to describe the data in terms of a density plot. The data and code is provided, and this is commendable. Additional comments: The research question posed is appropriate and the code is sufficient for the analysis of the same. However, the manuscript at this stage is not ready for publication and needs to be reworked for clarity. Several key points raised in section 2 in particular need to be addressed adequately. The paper contains implementation details and shows significant effort in terms of covering the existing metrics. I am certain it will be a suitable addition to the literature after revisions.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EXPERIMENTAL INTERPRETATION OF ADEQUATE WEIGHT-METRIC COMBINATION FOR DYNAMIC USER-BASED COLLABORATIVE FILTERING Review round: 2 Reviewer: 1

Basic reporting: The paper's readability has improved significantly. However, I did not intend to imply that the paper needed an editorial service; and I am sorry the authors had to retain one. Experimental design: N/A. Validity of the findings: N/A. Additional comments: The authors have addressed my previous comments; and the paper has been extensively reworked. At this point I am amenable towards this paper's acceptance.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EXPERIMENTAL INTERPRETATION OF ADEQUATE WEIGHT-METRIC COMBINATION FOR DYNAMIC USER-BASED COLLABORATIVE FILTERING Review round: 2 Reviewer: 2

Basic reporting: The structure of manuscript is not well-organized, and the content is not coherent, making it less readable. Experimental design: The procedure of the experiment and the settings of the parameters are described with sufficient detail. Meanwhile, the meaning of each metric is illustrated in detail. Validity of the findings: The dpi of the figures in the manuscript is too low. After downloading the figures from the attachment, some results can be verified. However, the reviewer cannot correspond to the ”BNC=10” in line 532, where did the authors get it? It should be emphasized that the discussion is difficult to be well understood, and it is necessary to report the results in a more structured manner. Additional comments: 1) The merits and demerits of the four touchstone similarity equations should be introduced in detail. 2) There are 8 formulas in the section of experimental design, the authors only analysed the PCC, why not discuss the other optimal BNCs? 3) On line 202 page 10, the way of citation is inconsistent with the previous content, please double-check. 4) Some abbreviations are defined repeatedly (e.g., PCC, MRC, COS, JAC in section ITEM INDEPENDENCY ANALYSES)

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EXPERIMENTAL INTERPRETATION OF ADEQUATE WEIGHT-METRIC COMBINATION FOR DYNAMIC USER-BASED COLLABORATIVE FILTERING Review round: 3 Reviewer: 1

Basic reporting: No comment. Experimental design: No comment. Validity of the findings: No comment. Additional comments: No comment.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EXPERIMENTAL INTERPRETATION OF ADEQUATE WEIGHT-METRIC COMBINATION FOR DYNAMIC USER-BASED COLLABORATIVE FILTERING Review round: 3 Reviewer: 2

Basic reporting: The structure and readability of the manuscript have been improved a lot. Experimental design: The procedure of the experiment and the settings of the parameters are described in sufficient detail. Validity of the findings: no comment Additional comments: The revised manuscript has been greatly improved and is suitable for publication.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: THE EXTERNAL AND DATA LOOSE COUPLING FOR THE INTEGRATION OF SOFTWARE UNITS: A SYSTEMATIC MAPPING STUDY Review round: 1 Reviewer: 1

Basic reporting: The paper presents a systematic mapping study (SMS) aimed at assessing the most recent (2008-2019) proposals, architectures, and technologies supporting the loosely coupled integration of software components, especially, in the field of Enterprise Application Integration (EAI). The paper is clear, and well organised in structure, despite the English quality could be definitely improved but letting a native speaker read the paper: there are a number of recurrent typos. A number of topics are surveyed in the paper, and for each of them, relevant entry point papers are referenced. However, the amount of information provided for each topic (e.g. in sec 2) is very concise, and barely sufficient for the reader which already has a basic understanding of the topic---therefore, I believe, more information should be provided for the readers which are not confident with those topics. Images and tables are of adequate quality, despite not all images are vectorial. Figures 7 and 8 are somewhat low quality. Maybe a listing would more adequate. I bet raw data could not be made available -- as it consists of the surveyed papers, I suppose -- yet, the authors share some intermediate data they have collected to produce the SMS, via a publicly available Google Sheet document. May I suggest sharing the data using a less volatile means? E.g. GitHub or Codeocean (maybe ask the editor what's the most adequate platform in their opinion). Surveys of this sort are very important within the scopes of computer science and software engineering, and they cover a quite large amount of aspects of these disciplines. Furthermore, I'm not aware of any similar survey since the last decade, so I believe the contribution is useful and timely. The introduction of the paper is clear, despite I struggle in understanding what does the "external" word -- which occurs into the title, abstract, and introduction -- refers to. Experimental design: In my opinion, the study design could be improved in a couple of ways. All the current research questions attempt to identify _which_ proposals/architectures/technologies are available into the literature to deal with loosely coupling and enterprise integration. For this reason, the results (sec. 4) are slightly more than an enumeration of keywords and references. This is for sure an extremely useful starting point, yet does not help the reader in understanding _how_ and to what extent the enumerated proposals/architectures/technologies favour loosely coupled integration. Maybe more queries and further analysis may be required to serve this purpose. Also, I am not sure the current study design could be capable of capturing a shift in the meaning of the terms used to select primary studies. In other words, it seems to me that the authors are tailoring their queries to some terms which were extensively used one decade ago. Therefore, it is unsurprising that they are able to find proposals/architectures/technologies from around that period. It seems to me that the current design of the study does a good job in understanding the evolution of technologies of EAI in the last decade, despite it may struggle in figuring out whether new proposals/architectures/technologies have joined the game in the meanwhile. Validity of the findings: At its current state, section 5 (discussion) is essentially useless. It spends a few pages discussing the rigidity of SOA approaches due to syntactical aspects of WDSL and... that's basically it. I believe a more elaborate discussion is needed. An interesting discussion, in my view, should attempt to investigate the actual contribution of the selected proposals/architectures/technologies w.r.t. the problem of software integration (i.e. how they solve it), thus eliciting which problems can be currently considered solved and, possibly, which ones cannot. Also, I believe the author should elaborate their conclusions (sec 6) better than this. Currently, the section consists of 3 paragraphs: the first one summarises the methodology, the second one essentially states that the major issue of integration is WDSL (ok, but it sounds somewhat reductive to me), and the last one briefly summarises the importance of EAI, its reliance on proprietary technologies (which has never been discussed before in the paper) and some future works involving "Dynamic Data Canonical Model" and "Agnostic Messages"---both aspects are lacking references and have not been described before in the paper. Additional comments: No additional comment.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: THE EXTERNAL AND DATA LOOSE COUPLING FOR THE INTEGRATION OF SOFTWARE UNITS: A SYSTEMATIC MAPPING STUDY Review round: 1 Reviewer: 2

Basic reporting: The paper tackles an interesting problem related to how software units interconnect and focuses on the concept and current body of knowledge related to loose coupling. It performs a systematic mapping study to analyze the existing challenges in this domain. The paper generally reads well, and the topic is very living to the journal. It is also essential for the software engineering community to be up to date with all the challenges exposed by software developers and researchers. I have a few comments about the study design and results, which I will enumerate as I review the paper: Experimental design: I did not understand why the authors have associated their research questions with the period of 2008 all the way to 2019. Also, it was not clear to me why the authors have chosen to focus and this period exactly, keeping in mind that most of the references that the authors have used mainly to introduce all the concepts that they are surveying were actually pretty old and, to say the least, are before 2008. So, the choice of the period needs to be justified. I think the authors should carefully explain how they performed the classification of the papers going from the selection of the potential venues to explore, all the way to the choice of documents using inclusion and exclusion criteria. Yet, it was not clear to me how the authors have moved from a total of 3,095 articles all the way to 95 articles. If this has been done using the procedure outlined in figure number two, then this would be just along for manual analysis to what extent were able to do it without any buttons without missing any studies properly. The authors have stated that the goal of the study is to identify gaps in current research to suggest areas for further investigation. However, the research questions outlined to fulfill this goal are most likely to be just exploratory and reporting numbers related to primary studies related to software unit integration and existing architectures and frameworks. Little is known about how exactly the Gap will be detected and how the authors have framed this Gap in terms of missing technology or with respect to the challenges related to software unit integration. In this context, I was expecting to see more limitations, errors, issues and problems related to each technology or framework and how existing studies are or not covering it. Validity of the findings: This is also reflected in the results and mainly the discussion section that likes deep analysis of the limitations of existing papers. All the reflections made in the discussion are rather generic and known. Even the example brought by the authors about the limitation of defining data types during design time then dealing with them at the runtime is already a known problem. Besides being already known, this limitation does not seem to be extracted directly from the studies but rather it is a reflection of the authors of what they think it is a problem in these papers. Section 4 which is called analysis of results seems to be just reporting results without any analysis. I did not see anything deeper than just specifying papers, which can be done using a classifier with proper training. The manual validation of the authors would have proposed a better understanding of these papers and what are the main challenges they are trying to solve. That would be better to improve the paper. Additional comments: The authors mentioned the use of an excel sheet for their PS. It would be strongly recommended to share it with the community for replication and extension purposes.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: THE EXTERNAL AND DATA LOOSE COUPLING FOR THE INTEGRATION OF SOFTWARE UNITS: A SYSTEMATIC MAPPING STUDY Review round: 2 Reviewer: 1

Basic reporting: I am fairly satisfied with the improvements the authors performed in their last revision. I believe my previous concerns have been reasonably addressed, and I have no further concerns. Experimental design: Nothing new w.r.t. my previous review concerning the study design. Validity of the findings: Nothing new w.r.t. my previous review concerning the validity of the findings. Additional comments: I still believe that figures 7 and 8, as well as their descriptions in sec. 4, are maybe too much detailed and purpose-specific for the scope of this paper. I believe the authors may consider generalising that discussion to languages other than XSD or XML. Also, there are still a few typos, e.g.: - line 115, double ")" - line 153, missing "JSON"

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: THE EXTERNAL AND DATA LOOSE COUPLING FOR THE INTEGRATION OF SOFTWARE UNITS: A SYSTEMATIC MAPPING STUDY Review round: 2 Reviewer: 2

Basic reporting: NA Experimental design: NA Validity of the findings: NA Additional comments: I would like to thank the authors for carefully addressing my comments. I no longer have any concerns with this paper.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: FRAMEWORKS FOR DEVELOPING AN AGRO-PROSUMER COMMUNITY GROUP PLATFORM Review round: 1 Reviewer: 1

Basic reporting: - The paper proposes three frameworks for Agro Prosumer Community Group. - The first framework defines the prosumers and prerequisites. Methodology of the proposed framework is based on clustering prosumer profiles for detection of outliers and optimization of prosumer clusters. - The novelty aspect of the paper is developing a method for recruitment of new Agro prosumer based on real-time production profile over a period of two years (evaluation period) rather than relying on historic production behavior. In this framework, authors have identified methods and developed an evaluation criteria for recruitment of eligible prosumers. Authors used MATLAB platform to test the evaluation criteria. - In addition, authors also used a goal-oriented approach for the development of goal management framework 3 for Agro Prosumer Community Group. Authors used goal programming language for multiple conflicting goals to prioritize the goals and optimise the goal management framework using LINGO-2. There are few suggestions: - Equation number 3 is missing in the paper. - There are only few references in the paper. - In recruitment framework, authors mentioned that the recruitment of prosumers assumes that they cannot change the selection of preferred APCG. There is no discussion and hence it remains unclear how fair this assumption is. - Also, the verification and data generation process described in framework 2 is not very clear. Authors are advised to add more details in this section. Experimental design: - Research questions are framed properly. - Authors have presented multiple frameworks to address the key research questions. - Knowledge gaps are identified properly and proposed frameworks fill the gaps. Validity of the findings: - In recruitment framework, authors mentioned that the recruitment of prosumers assumes that they cannot change the selection of preferred APCG. There is no discussion and hence it remains unclear how fair this assumption is. - Also, the verification and data generation process described in framework 2 is not very clear. Authors are advised to add more details in this section. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: FRAMEWORKS FOR DEVELOPING AN AGRO-PROSUMER COMMUNITY GROUP PLATFORM Review round: 1 Reviewer: 2

Basic reporting: no comment Experimental design: no comment Validity of the findings: no comment Additional comments: no comment

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 1 Reviewer: 1

Basic reporting: Typos and English errors Experimental design: no comment Validity of the findings: no comment Additional comments: What is the base of author claiming that DPWSS my be the first private working set selection algorithm based on differential privacy? Abstract is well-written but can include more info about experimental results. Related work section is too short. Authors should include some recent works and list their limitations. Authors can include references such as: A Review of Machine Learning Algorithms for Cloud Computing Security Performance Assessment of Supervised Classifiers for Designing Intrusion Detection Systems: A Comprehensive Review and Recommendations for Future Research Author Classification using Transfer Learning and Predicting Stars in Co-Author Networks Unnecessary background has increased the paper length. Can you compare your approach with an existing approach. Pl revise conclusion to conclude the work. Pl proof-read the manuscript.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 1 Reviewer: 2

Basic reporting: Authors propose a new differentially private working set selection algorithm (DPWSS), which utilizes exponential mechanism to privately select working sets. To the best of our knowledge, DPWSS is the first private working set selection algorithm based on differential privacy. The authors theoretically prove that the proposed algorithm satisfies differential privacy. The extended experiments show that DPWSS can achieve classification accuracy and optimized objective value almost the same as the original non-privacy SVM algorithm with higher execution efficiency. Experimental design: Authors propose a new differentially private working set selection algorithm (DPWSS), which utilizes exponential mechanism to privately select working sets. To the best of our knowledge, DPWSS is the first private working set selection algorithm based on differential privacy. The authors theoretically prove that the proposed algorithm satisfies differential privacy. The extended experiments show that DPWSS can achieve classification accuracy and optimized objective value almost the same as the original non-privacy SVM algorithm with higher execution efficiency. Validity of the findings: Authors propose a new differentially private working set selection algorithm (DPWSS), which utilizes exponential mechanism to privately select working sets. To the best of our knowledge, DPWSS is the first private working set selection algorithm based on differential privacy. The authors theoretically prove that the proposed algorithm satisfies differential privacy. The extended experiments show that DPWSS can achieve classification accuracy and optimized objective value almost the same as the original non-privacy SVM algorithm with higher execution efficiency. Additional comments: 1. Please improve overall readability of the paper. 2. The objectives of this paper need to be polished. 3. Introduction is poorly written. 4. Relevant literature review of latest similar research studies on the topic at hand must be discussed. 5. Result section need to be polished. 6. There are some grammar and typo errors. 7. Improve the quality of figures 8. Define all the variables before using The authors can cite the following refrences 1. A Novel PCA-Firefly based XGBoost classification model for Intrusion Detection in Networks using GPU 2. A Novel PCA-Firefly based XGBoost classification model for Intrusion Detection in Networks using GPU

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 1 Reviewer: 3

Basic reporting: In this paper, the authors proposed a new differentially private working set selection algorithm (DPWSS), which utilizes an exponential mechanism to privately select working sets. This paper is providing a good contribution to the state of the art. Experimental design: The extended experiments show that DPWSS can achieve classification accuracy and optimized objective value almost the same as the original non-privacy SVM algorithm with higher execution efficiency. Validity of the findings: To the best of our knowledge, DPWSS is the first private working set selection algorithm based on differential privacy Additional comments: Please improve the abstract. It should highlight the background as well and add some factual results that shows how much results have your approach achieved. 2. Introduction is too long. Please reduce it. It should be specific. 3. in introduction.... Support vector machine (SVM)...... quadratic programming (QP......... Sequential Minimal Optimization (SMO).... make it in a single format. 4. The quality of the figures can be improved more. 5. Result section is written badly. Explain results extensively and clearly explain the flow of results. 6. All equations, tables and figure should be cited in the text. 7. The author should add more details of the results and improve the mathematical models. 8. Future work and Challenges are the important things that lead the researchers to take the current state-of-the-art to further step. The authors should explain What open issues and challenges are needed to be addressed by the researchers in this field? 9. What are the evaluations used for the verification of results (like, roc curve)? 10. Abbreviations and Acronyms should be added only the first time. Then only acronyms should be used in the entire paper. 11. Few latest references can be added and explained a bit in the related work section. some are given below 12. Problem statement must be in the introduction. 13. Grammar check is required. Fire detection method based on improved fruit fly optimization-based SVM A Novel PCA-Firefly based XGBoost classification model for Intrusion Detection in Networks using GPU Privacy protection algorithm for the internet of vehicles based on local differential privacy and game model Smo dnn Spider monkey optimization and deep neural network hybrid classifier model for intrusion detection Frequent itemset mining of user’s multi-attribute under local differential privacy

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 2 Reviewer: 1

Basic reporting: Proof-reading is still required. Experimental design: NA Validity of the findings: NA Additional comments: Related work section is still too short. Authors didn't address the comment. Authors should include some recent works and list their limitations. Authors can include references such as: A Consolidated Decision Tree-based Intrusion Detection System for binary and multiclass imbalanced datasets Performance Assessment of Supervised Classifiers for Designing Intrusion Detection Systems: A Comprehensive Review and Recommendations for Future Research Conclusion can be improved by adding limitations of the proposed work. Proof-reading is required.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 2 Reviewer: 2

Basic reporting: The authors have addressed all of my comments, paper can be accepted in the current form. Thank you for giving me this opportunity. Experimental design: Good Validity of the findings: Good Additional comments: The authors have addressed all of my comments, paper can be accepted in the current form. Thank you for giving me this opportunity.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 2 Reviewer: 3

Basic reporting: no comment Experimental design: no comment Validity of the findings: no comment Additional comments: authors have addressed most of the comments, but need to revised all comments suggested. Equations 10 and 11 should be described the significance. in section 4.5 privacy analysis one paragraph is needed to describe the process in text. suggested related references are not cited yet.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 3 Reviewer: 1

Basic reporting: In this paper, the authors tackled an interesting problem of selecting a training set for support vector machine while maintaining privacy of the data. The topic is worthy of investigation, but the following issues need to be thoroughly addressed before the manuscript could be considered for publication: 1. There are quite a number of vague statements across the manuscript. As an example, what do the authors mean by saying (in the abstract): “The errors of optimized objective value between the two algorithms were nearly less than two”? And then, “the DPWSS algorithm had a higher execution efficiency by comparing iterations on different datasets” – higher than what? Than the “original non-privacy SVM”? 2. Please avoid using bulked references, such as “Support vector machine (SVM)[1][2][3][4][5]” – it does not bring any useful to the reader. Please unfold such bulked references. 3. The authors should clearly state what do they mean by the “working set selection”. 4. Please avoid naming sections as “DP” etc (I suggest unfolding acronyms in the section/subsection headers). 5. What is the time and memory complexity of the proposed algorithm? 6. The authors sort of failed to contextualize their work within the state of the art of training SVMs. There have been a multitude of methods proposed so far, which could potentially impact maintaining privacy of the classifier. As an example, there are a number of training set selection methods around which could be used for selecting such training examples that could help maintain privacy. Are such methods possible to combine with the proposed approach? It would be useful to discuss that to elaborate a bit more comprehensive view of the problem of SVM training. For some references on training set selection for SVMs, the authors may want to check e.g., https://link.springer.com/article/10.1007/s10462-017-9611-1, https://link.springer.com/chapter/10.1007/11539087_71. 7. Although the English is acceptable, the manuscript would benefit from proofreading (there are several grammatical errors around). Experimental design: 1. The characteristics of the datasets used in the experimental study must be expanded. Specifically, what is the imbalance ratio within each dataset? Does it have any impact on the overall performance of the optimizer? Also, the number of datasets is fairly small – currently, it is common to test such SVM-related algorithms over 50+ datasets to fully understand their generalization abilities. To this end, I suggest including much more benchmarks in the experimental validation (especially given that only one dataset may be considered “large”). 2. Please present all quality metrics used in the manuscript in a formal (mathematical way) through providing their formulae. 3. It would be useful to at least discuss if the proposed method is suitable for multi-class classification using support vector machines (e.g., in the one-vs-all strategy). 4. The authors should present all relevant quality metrics: AUC, accuracy, precision, recall, F1 and MCC. 5. The authors should back up their conclusions with appropriate statistical testing (here, non-parametric statistical tests) to fully understand if the differences between different optimizers are significant in the statistical sense. Validity of the findings: In my opinion, the authors should extend the experimental validation (as suggested in my previous comments), to provide clear and unbiased view on the generalization abilities of the proposed techniques. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 3 Reviewer: 2

Basic reporting: This paper presents a differentially private working set selection to train support vector machine. They focus on decomposition in the sequential minimal optimization and perturb each iteration with exponential mechanism. The novelty is quite limited to this optimization and no empirical comparison to any existing differentially private SVM. The paper is clearly organized while some of the terms and notations are not consistently explained, such as constant-factor. I also have major concerns for the illustration of the results and evaluation design as discussed below. Experimental design: I notice that the authors emphasize that their scheme is free from the influence of high-dimensional data on noise while their evaluation is mostly on dataset with at most a few hundred dimensions. The author should clarify the definition of “high-dimension” in the case of differential privacy adoption in SVM and address this claim by using a substantially high-dimensional dataset. Validity of the findings: For Figure 2 to 9, it is very challenging to tell which scheme is superior as they look the same to me. I highly suggest the authors to start the y-axis at a higher value such as 0.6. Similar issues are on Figure 10-12 given none of the combination gives an error over 2.5. The authors can scale up the value to address the performance difference. Also it is confusing to use curve chart when plotting the results for different \epsilon and \delta combination as they do not represent any sort of performance trend. If the authors intend to demonstrate the impact of different parameters, they should fix on one and vary the other. Two graphs or matrix tables are more suitable for that purpose. Additional comments: Minor: Section 3.1: for training a SVM -> an Section 5.3: under the circumstances most of the violating -> under the circumstances, most

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 4 Reviewer: 1

Basic reporting: Unfortunately, although the authors tried to address my concerns, the most important ones still remain unresolved. Specifically, please refer to the following comments from my previous review: 2), 3), 7). Experimental design: The authors should still work on the experimental validation of the proposed algorithm. Specifically, the authors should still expand their benchmarks, present more detailed experimental results captured using the metrics I mentioned in my previous review, and should present clear statistical evidence (backed up with appropriate statistical testing) behind the claims. I consider the following comments from my previous review still unresolved: 1), 4), 5). Validity of the findings: See my previous comments Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DPWSS: DIFFERENTIALLY PRIVATE WORKING SET SELECTION FOR TRAINING SUPPORT VECTOR MACHINES Review round: 5 Reviewer: 1

Basic reporting: I am happy to see that the authors have addressed my concerns, and the manuscript is in better shape now. Experimental design: - Validity of the findings: - Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: AN EFFICIENT NETWORKING SOLUTION FOR EXTENDING AND CONTROLLING WIRELESS SENSOR NETWORKS USING LOW-ENERGY TECHNOLOGIES Review round: 1 Reviewer: 1

Basic reporting: no comment Experimental design: no comment Validity of the findings: no comment Additional comments: The authors proposed proposes an approach for constructing and extending Bi-Directional mesh networks using low power consumption technologies inside various indoors and outdoors architectures called "an adaptable Spider - Mesh topology.". The use of ESP-NOW protocol as a communication technology added the advantage of longer communication distance versus a slight increase of consumed power. It provides 15 times longer distance compared to the BLE protocol while consuming only twice as much power. In my opinion, this work is interesting and should be published after minor revisions. I consider that the authors should clarify several aspects of the manuscript, for improving the clarity of the presentation. These are listed below. 1. In particular, the term running current is used in Electric Motor Testing, why authors have used the term in the manuscript? 2. The authors can be to consider show their lifetime experimental results in a table. 3. Can the authors measure the time of the response responses time between the different topologies?

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: AN EFFICIENT NETWORKING SOLUTION FOR EXTENDING AND CONTROLLING WIRELESS SENSOR NETWORKS USING LOW-ENERGY TECHNOLOGIES Review round: 1 Reviewer: 2

Basic reporting: This paper proposes an energy-efficient routing protocol for wireless sensor networks. The paper is well-written and easy to follow. Following are my comments: 1. There has been a great amount of work done on developing energy-efficient routing protocols, it would be better to distinguish your work from the existing literature or clearly mention the novel contributions. 2. I wonder if the proposed work can be used in extreme environments such as in-body or underwater {https://repository.kaust.edu.sa/handle/10754/664913, https://ieeexplore.ieee.org/abstract/document/8891506} 3. The results do not show a comparison to existing works to show its effectiveness. Experimental design: The experimental design is clear. Validity of the findings: There is no issue with the findings, only comparisons to existing schemes is missing. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: AN EFFICIENT NETWORKING SOLUTION FOR EXTENDING AND CONTROLLING WIRELESS SENSOR NETWORKS USING LOW-ENERGY TECHNOLOGIES Review round: 2 Reviewer: 1

Basic reporting: no comment Experimental design: no comment Validity of the findings: no comment Additional comments: I believe that the manuscript is now ready to be published in PeerJ

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: AN EFFICIENT NETWORKING SOLUTION FOR EXTENDING AND CONTROLLING WIRELESS SENSOR NETWORKS USING LOW-ENERGY TECHNOLOGIES Review round: 2 Reviewer: 2

Basic reporting: The authors have addressed all the comments. No further comments from this reviewer. Experimental design: Experimental design is fine. Validity of the findings: The findings are valid. Additional comments: The authors have addressed all the comments. No further comments from this reviewer.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DETECTING SECURITY ATTACKS IN CYBER-PHYSICAL SYSTEMS: A COMPARISON OF MULE AND WSO2 INTELLIGENT IOT ARCHITECTURES Review round: 1 Reviewer: 1

Basic reporting: Paper theme is good and it will attract attention in community. Experimental design: Experimental process is explained in proper way. Validity of the findings: Validation is done with help of comparative analysis. Additional comments: 1. There is no link of code in the paper. Add github or any other relevant link so that proposed work may be tested. 2. Highlight all assumptions and limitations of your work. 3. Mention all figures properly in text. 4. Mention time complexity of proposed pipeline.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DETECTING SECURITY ATTACKS IN CYBER-PHYSICAL SYSTEMS: A COMPARISON OF MULE AND WSO2 INTELLIGENT IOT ARCHITECTURES Review round: 1 Reviewer: 2

Basic reporting: This review is for the paper to be published in PeerJ Computer Science #61593. The title of the manuscript is: “Detecting security attacks in cyber-physical systems: a comparison of mule and WSO2 intelligent IoT architectures”, written by José Roldán-Gómez, Juan Boubeta-Puig, Gabriela Pachacama-Castillo, Guadalupe Ortiz, José Luis Martínez. The authors wrote the article in good professional English. As the reader, I can easily follow the storyline with ease and clarity. I believe that the article conforms to scientific standards and follows courtesy and expressions. In the introduction in lines 36,37, I am afraid I have to disagree with the authors about the lack of discussion in the IoT securities space. In my opinion, IoT securities have been discussed and published extensively. The authors should provide more evidence to convince the reader about this statement and how this research contributes novelty to the IoT securities domain. The literature review is unusually described towards the paper's final section, which is after the results and discussions(lines 661). Usually, the article's structure would have a prior literature review conducted to determine the proper methodology for this purpose. However, the literature review being presented is to compare the performance analysis of this research. I understand what the authors intend to do. By all means, it is not a wrong thing to do, but I would like to suggest the authors should also provide a literature review in order to support the chosen research method. For example, when the authors compare the architecture implementation with Mule and Esper versus WSO2 and Siddhi, the authors should provide broader literature to convince the reader that the architecture being discussed is better than other architecture. In line 343, the author should justify the use of the MQTT network simulator. How is this simulator a better choice compared to other network simulators? In my opinion, the article already includes an adequate introduction. The background section demonstrates the broader space of knowledge that fits into this research. The authors should provide more initial relevant literature to support this research. For example, in lines 155,156 about the TCP and UDP flood attacks, the authors should provide more evidence to support the statement. There is no issue with the structure of the article. Some formats in this paper may not be standard, but I can follow the paper with sufficient clarity as a reader. Figures and tables are also written in a good format. The paper suggests 4 research questions, and the results are comprehensive and sufficiently self-contained. The report includes all results coherently relevant to the research questions. Formal results have already included clear definitions of all terms and theorems, detailed proofs, and measurements. Experimental design: I believe that this paper is original, and the primary research fits within the aim and scope of the PeerJ Computer Science journal. The research question is sufficiently well defined, relevant & meaningful. The research fills an identified knowledge gap. The submission should clearly define the research question, which must be relevant and meaningful. The knowledge gap being investigated should be identified, and statements about how the study contributes to filling that gap. The investigation has been conducted rigorously, and it conforms to a high technical standard. I would also state that the research is conducted according to ethical standards, particularly in the IoT Security Space. Validity of the findings: Impact and novelty are sufficiently assessed. The methods described do not provide sufficient information to be reproducible by another investigator. Unless the authors provide the complete programmable codes, then other researchers could replicate these findings. All underlying data are sufficiently robust, statistically sound, and well-controlled. Conclusions are sufficiently well stated and linked to the original research question. I would like to comment on lines 93-102; in conclusion, I am not convinced whether this aim has already been achieved. Should this aim is successfully achieved, how would it substantially benefit humanity. In general, the conclusions are appropriately stated, connected to the original question investigated, and supported the results with clarity. A well-controlled experimental intervention sufficiently supports claims of a causative relationship. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DETECTING SECURITY ATTACKS IN CYBER-PHYSICAL SYSTEMS: A COMPARISON OF MULE AND WSO2 INTELLIGENT IOT ARCHITECTURES Review round: 1 Reviewer: 3

Basic reporting: The authors presented their work entitled, Detecting Security Attacks in 2 Cyber-Physical Systems: A Comparison of 3 Mule and WSO2 Intelligent IoT Architectures. Where authors claims that The Internet of Things (IoT) paradigm keeps growing, and many different IoT devices, such as smartphones and smart appliances, are extensively used in smart industries and smart cities. The benefits of this paradigm are obvious, but these IoT environments have brought with them new challenges, such as detecting and combating cybersecurity attacks against cyber-physical systems. This paper addresses the real-time detection of security attacks in these IoT systems through the combined used of Machine Learning (ML) techniques and Complex Event Processing (CEP). In this regard, in the past we proposed an intelligent architecture that integrates ML with CEP, and which permits the definition of event patterns for the real-time detection of not only specific IoT security attacks, but also novel attacks that have not previously been defined. Our current concern, and the main objective of this paper, is to ensure that the architecture is not necessarily linked to specific vendor technologies and that it can be implemented with other vendor technologies while maintaining its correct functionality. We also set out to evaluate and compare the performance and benefits of alternative implementations. This is why the proposed architecture has been implemented by using technologies from different vendors: firstly, the Mule Enterprise Service Bus (ESB) together with the Esper CEP engine; and secondly, the WSO2 ESB with the Siddhi CEP engine. Both implementations have been tested in terms of performance and stress, and they are compared and discussed in this paper. The results obtained demonstrate that both implementations are suitable and effective. However, I can see great room for improvement in multiple sections. 1. The abstract does not align with the paper actual findings and conclusion as well. 2. Figures 2, 3 and a few more are not readable clearly 3. Too many figures, while elaboration is not extensive, authors may elaborate properly. 4. Conclusion requires revision, it can be concise with the findings only, currently, it is too lengthy. 5. Proofread is highly recommended. 6. May elaborate more on their proposed approach. 7. Provided references are better enough. However, a few of them are missing information, and can be strengthen further. Experimental design: The authors presented their work entitled, Detecting Security Attacks in 2 Cyber-Physical Systems: A Comparison of 3 Mule and WSO2 Intelligent IoT Architectures. Where authors claims that The Internet of Things (IoT) paradigm keeps growing, and many different IoT devices, such as smartphones and smart appliances, are extensively used in smart industries and smart cities. The benefits of this paradigm are obvious, but these IoT environments have brought with them new challenges, such as detecting and combating cybersecurity attacks against cyber-physical systems. This paper addresses the real-time detection of security attacks in these IoT systems through the combined used of Machine Learning (ML) techniques and Complex Event Processing (CEP). In this regard, in the past we proposed an intelligent architecture that integrates ML with CEP, and which permits the definition of event patterns for the real-time detection of not only specific IoT security attacks, but also novel attacks that have not previously been defined. Our current concern, and the main objective of this paper, is to ensure that the architecture is not necessarily linked to specific vendor technologies and that it can be implemented with other vendor technologies while maintaining its correct functionality. We also set out to evaluate and compare the performance and benefits of alternative implementations. This is why the proposed architecture has been implemented by using technologies from different vendors: firstly, the Mule Enterprise Service Bus (ESB) together with the Esper CEP engine; and secondly, the WSO2 ESB with the Siddhi CEP engine. Both implementations have been tested in terms of performance and stress, and they are compared and discussed in this paper. The results obtained demonstrate that both implementations are suitable and effective. However, I can see great room for improvement in multiple sections. 1. The abstract does not align with the paper actual findings and conclusion as well. 2. Figures 2, 3 and a few more are not readable clearly 3. Too many figures, while elaboration is not extensive, authors may elaborate properly. 4. Conclusion requires revision, it can be concise with the findings only, currently, it is too lengthy. 5. Proofread is highly recommended. 6. May elaborate more on their proposed approach. 7. Provided references are better enough. However, a few of them are missing information, and can be strengthen further. Validity of the findings: The authors presented their work entitled, Detecting Security Attacks in 2 Cyber-Physical Systems: A Comparison of 3 Mule and WSO2 Intelligent IoT Architectures. Where authors claims that The Internet of Things (IoT) paradigm keeps growing, and many different IoT devices, such as smartphones and smart appliances, are extensively used in smart industries and smart cities. The benefits of this paradigm are obvious, but these IoT environments have brought with them new challenges, such as detecting and combating cybersecurity attacks against cyber-physical systems. This paper addresses the real-time detection of security attacks in these IoT systems through the combined used of Machine Learning (ML) techniques and Complex Event Processing (CEP). In this regard, in the past we proposed an intelligent architecture that integrates ML with CEP, and which permits the definition of event patterns for the real-time detection of not only specific IoT security attacks, but also novel attacks that have not previously been defined. Our current concern, and the main objective of this paper, is to ensure that the architecture is not necessarily linked to specific vendor technologies and that it can be implemented with other vendor technologies while maintaining its correct functionality. We also set out to evaluate and compare the performance and benefits of alternative implementations. This is why the proposed architecture has been implemented by using technologies from different vendors: firstly, the Mule Enterprise Service Bus (ESB) together with the Esper CEP engine; and secondly, the WSO2 ESB with the Siddhi CEP engine. Both implementations have been tested in terms of performance and stress, and they are compared and discussed in this paper. The results obtained demonstrate that both implementations are suitable and effective. However, I can see great room for improvement in multiple sections. 1. The abstract does not align with the paper actual findings and conclusion as well. 2. Figures 2, 3 and a few more are not readable clearly 3. Too many figures, while elaboration is not extensive, authors may elaborate properly. 4. Conclusion requires revision, it can be concise with the findings only, currently, it is too lengthy. 5. Proofread is highly recommended. 6. May elaborate more on their proposed approach. 7. Provided references are better enough. However, a few of them are missing information, and can be strengthen further. Additional comments: The authors presented their work entitled, Detecting Security Attacks in 2 Cyber-Physical Systems: A Comparison of 3 Mule and WSO2 Intelligent IoT Architectures. Where authors claims that The Internet of Things (IoT) paradigm keeps growing, and many different IoT devices, such as smartphones and smart appliances, are extensively used in smart industries and smart cities. The benefits of this paradigm are obvious, but these IoT environments have brought with them new challenges, such as detecting and combating cybersecurity attacks against cyber-physical systems. This paper addresses the real-time detection of security attacks in these IoT systems through the combined used of Machine Learning (ML) techniques and Complex Event Processing (CEP). In this regard, in the past we proposed an intelligent architecture that integrates ML with CEP, and which permits the definition of event patterns for the real-time detection of not only specific IoT security attacks, but also novel attacks that have not previously been defined. Our current concern, and the main objective of this paper, is to ensure that the architecture is not necessarily linked to specific vendor technologies and that it can be implemented with other vendor technologies while maintaining its correct functionality. We also set out to evaluate and compare the performance and benefits of alternative implementations. This is why the proposed architecture has been implemented by using technologies from different vendors: firstly, the Mule Enterprise Service Bus (ESB) together with the Esper CEP engine; and secondly, the WSO2 ESB with the Siddhi CEP engine. Both implementations have been tested in terms of performance and stress, and they are compared and discussed in this paper. The results obtained demonstrate that both implementations are suitable and effective. However, I can see great room for improvement in multiple sections. 1. The abstract does not align with the paper actual findings and conclusion as well. 2. Figures 2, 3 and a few more are not readable clearly 3. Too many figures, while elaboration is not extensive, authors may elaborate properly. 4. Conclusion requires revision, it can be concise with the findings only, currently, it is too lengthy. 5. Proofread is highly recommended. 6. May elaborate more on their proposed approach. 7. Provided references are better enough. However, a few of them are missing information, and can be strengthen further.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DETECTING SECURITY ATTACKS IN CYBER-PHYSICAL SYSTEMS: A COMPARISON OF MULE AND WSO2 INTELLIGENT IOT ARCHITECTURES Review round: 2 Reviewer: 1

Basic reporting: All the changes have been done as per suggestions of reviewers. Current form of paper is suitable for publication. Experimental design: Fine Validity of the findings: Fine Additional comments: NA

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DETECTING SECURITY ATTACKS IN CYBER-PHYSICAL SYSTEMS: A COMPARISON OF MULE AND WSO2 INTELLIGENT IOT ARCHITECTURES Review round: 2 Reviewer: 2

Basic reporting: The authors wrote the article in good professional English. As the reader, I can easily follow the storyline with ease and clarity. There is no issue with the structure of the article. Some formats in this paper may not be standard, but I can follow the paper with sufficient clarity as a reader. Figures and tables are also written in a good format. Experimental design: The paper contains relevant information adequate to justify publication, demonstrating a proper understanding of the relevant literature. No considerable work is ignored. The paper's argument is already built on an appropriate base of theory, concepts, and other ideas. The research or equivalent intellectual work on which the article is already based has been well designed. And the methods have been employed appropriately. Validity of the findings: The results are presented clearly and analyzed appropriately, and the conclusions adequately tie together the other elements of the paper. The paper already identifies any implications practice and the implications consistent with the findings and conclusions of the article.  The article clearly expresses its case, measured against the technical language of the field. I would be happy if this paper could be published in this journal. A reasonable effort has been made. By reviewing this paper, I gained more insight into IoT, an essential application for blockchain technology that I am working on. Additional comments: To begin with, I would like to thank the authors for considering the Peer-J Computer Science as a publication outlet for their paper "Detecting Security Attacks in 2 Cyber-Physical Systems: A Comparison of 3 Mule and WSO2 Intelligent IoT Architectures". The report addresses the real-time detection of security attacks in these IoT systems through the combined use of Machine Learning (ML) techniques and Complex Event Processing (CEP). The paper has improved in comparison to its previous version. Overall this paper is a clear, concise, and well-written manuscript. The introduction is relevant and theory-based. Sufficient information about the previous study findings is presented for readers to follow the present study rationale and procedures. The methods are generally appropriate. Good luck.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: DETECTING SECURITY ATTACKS IN CYBER-PHYSICAL SYSTEMS: A COMPARISON OF MULE AND WSO2 INTELLIGENT IOT ARCHITECTURES Review round: 2 Reviewer: 3

Basic reporting: The authors address all the comments and concerns. The paper stands for acceptance now. Experimental design: The authors address all the comments and concerns. The paper stands for acceptance now. Validity of the findings: The authors address all the comments and concerns. The paper stands for acceptance now. Additional comments: The authors address all the comments and concerns. The paper stands for acceptance now.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: MAC PROTOCOL WITH GROUPING AWARENESS GMAC FOR LARGE SCALE INTERNET-OF-THINGS NETWORK Review round: 1 Reviewer: 1

Basic reporting: The Research Paper is Ok. And novel protocol is being proposed, but the paper needs more technical revisions and is subject for re-review. Literature review, and the proposed work needs more technical structure information to be elaborated. Experimental design: Experimental design is Ok. But needs more technical layout. Validity of the findings: Results are Ok. But analysis part is weak. And needs to be discussed with more detailed description. Additional comments: The Paper needs the following Strong Revisions and is subject for re-review, and after re-review, the final decision for the paper will be taken: 1. Introduction should be added with more information with regard to the Problem Definition, scope and even with some Background. Add Objectives of the paper at the end of Introduction. Add Organization of the paper. 2. Min 15-25 papers should be cited under literature review, and every paper should be elaborated with what is being observed, what is the novelty aspect and what experimental results are observed. At the end of literature review, highlight what overall technical gaps are observed that led to the design of the proposed methodology. 3. Give a Novel name of the proposed protocol---And add this in Title, Abstract and other sections uniformly in the paper. And under proposed protocol- Give the heading of the proposed protocol--Start with System Model. Then highlight the Architecture- Give Steps, Flowchart and Algorithm of the proposed protocol. 4. Under experimentation- Add Simulation Parameters in Table. Add assumptions of simulation. Give the detailed description of every point of testing under seperate head. Give the Data based table of the values of experimentation. 5. Add some Analysis section and do performance evaluation with more existing proposed protocols/. 6. Add future scope to this paper. 7. Add some more latest references to the paper.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: MAC PROTOCOL WITH GROUPING AWARENESS GMAC FOR LARGE SCALE INTERNET-OF-THINGS NETWORK Review round: 1 Reviewer: 2

Basic reporting: Topic selection is good and basic impression of paper falls positive. Experimental design: Proposed algorithm must be properly explained in text. Validity of the findings: Fine Additional comments: Highlight all assumptions and limitations of your work. Mention time complexity of entire pipeline. Mention all figures properly in text.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: MAC PROTOCOL WITH GROUPING AWARENESS GMAC FOR LARGE SCALE INTERNET-OF-THINGS NETWORK Review round: 1 Reviewer: 3

Basic reporting: Professional Proofread and English are required. References and literature review is sufficient. Experimental design: The rationale, aim, and objectives may need to be clearly defined in the Abstract and other sections. Validity of the findings: Consistency is missing in the paper. Conclusion, Limitation, and Future scope needs revision for more clarity and enhancing the readability of the paper. Additional comments: The authors are working on an important area of research. Few observations: 1. The main rationale/objective of the paper needs to be added in the abstract and Introduction. 2. The writing is ambiguous and more clarity is required to enhance the readability of the paper. 3. Consistency is missing in the paper. Conclusion, Limitation, and Future scope needs revision and elaborations for more clarity and enhancing the readability of the paper. 4. Mathematical Equations and paper require thorough proofread. 5. Figure 1 is talking about its application in healthcare, though in the paper authors mentioned having its applications in various domains, then rationale of providing only for healthcare is missing.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: MAC PROTOCOL WITH GROUPING AWARENESS GMAC FOR LARGE SCALE INTERNET-OF-THINGS NETWORK Review round: 2 Reviewer: 1

Basic reporting: The Revised Paper has presented the Literature review, structure and results in better manner. Experimental design: Yes, the results are satisfactory. Validity of the findings: The findings are Ok. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: MAC PROTOCOL WITH GROUPING AWARENESS GMAC FOR LARGE SCALE INTERNET-OF-THINGS NETWORK Review round: 2 Reviewer: 2

Basic reporting: All the changes have been done. Current form of paper is suitable for publication. Experimental design: All the changes have been done. Current form of paper is suitable for publication. Validity of the findings: All the changes have been done. Current form of paper is suitable for publication. Additional comments: Accepted

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: MAC PROTOCOL WITH GROUPING AWARENESS GMAC FOR LARGE SCALE INTERNET-OF-THINGS NETWORK Review round: 2 Reviewer: 3

Basic reporting: OK Experimental design: OK Validity of the findings: OK Additional comments: Suggested changes are incorporated. We may proceed.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A SYSTEMATIC LITERATURE REVIEW ON THE USABILITY OF MOBILE APPLICATIONS FOR VISUALLY IMPAIRED USERS Review round: 1 Reviewer: 1

Basic reporting: The review is of interest and within the scope of the journal. The documents presents a A Systematic Literature Review on the Usability of Mobile Applications for Visually Impaired Users. I know of a recent review about this topic but the document is in Spanish and is specifically about indoor technologies (View publication at https://latamt.ieeer9.org/index.php/transactions/article/view/4454). I suggest improving the abstract at lines 23- 24 to provide more justification for your study (specifically, you should expand upon the knowledge gap being filled). Other additional comments are: How does this figure (line 63) illustrate the classification? You are repeating the same classification information you already mentioned in here and added a % to each theme. What do you mean by "concludes the paper" in line 68. I mean, What is "that section: that concludes the paper? Is this figure in line 85 really necessary? You already mentioned the stages on the text. Besides, you don't mention the figure. Figures should not be placed without having been mentioned in the text. What do you mean by relevant? This is also mentioned in Table 2 but you need to be more specific about this unique inclusion/exclusion criteria (lines 135, 157, 158). Same comments/questions as before: Is this figure necessary? (Line 136) You already mentioned the number of papers per database in the text. Besides, you did not mention the figure. Figures should not be placed without having been mentioned in the text. At line 138- 139 It should say sub stage. I suggest to be more specific. Please rewrite this sentence as: "Therefore, importing citations into MENDELEY was mandatory in order to eliminate the duplicates. About figure 4: Same comment. This figure is not mentioned in the text. Figure is not a summary of just the stage mentioned, is a summary of all stages, hence, it should be mentioned before, as indicated. Figures should be placed the closest AFTER after mentioning them the first time. Please rewrite lines 207 and 208 as: The PRISMA diagram shown in Figure 7 illustrates all systematic literature processes used in this study. Be more specific in line 211. All researchers involved in this SLR... Experimental design: The survey methodology is ver consistent and well explained. However, I perceived a language-biased coverage of the subject. Why just documents in English? You need to add some justification/explanation. Validity of the findings: The paper illustrates the different trends, themes, and evaluation results of various mobile applications developed in the last six years. Using this overview as a foundation, future directions for research in the field of usability for the visually impaired (UVI) are highlighted. More discussion about those futures directions needs to be added. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A SYSTEMATIC LITERATURE REVIEW ON THE USABILITY OF MOBILE APPLICATIONS FOR VISUALLY IMPAIRED USERS Review round: 1 Reviewer: 2

Basic reporting: The paper presents a systematic literature review on the usability of mobile applications for visually impaired users, covering a synthesis of 60 studies selected from an initial sample of 932 studies encountered in the literature from the last six years. My first suggestion for improvements is to include more details of the findings and type of knowledge synthesized in the study in the abstract. The information on results and their impact is very briefly described in the abstract, which would jeopardize the chances of reader having a broader view of what the paper has to offer before reading the full text. There are very long paragraphs in the text. The first paragraph of the introduction is very long, and I would strongly recommend the authors break it down into shorter paragraphs. The introduction has definitions to key terms, such as usability, with no reference to authoritative sources. I also believe it is important that the authors highlight the research gap they identified and that motivated the present study. Have other studies performed systematic reviews on the usability of mobile applications for visually impaired people? Why did the authors choose the period of six years? Line 52: fix citation style: evaluation tests. (Bastien, 2010). Experimental design: The search protocol should describe the different adaptations the search string had to undergo to be executed in each database. The quality assessment, starting on line 180, does not describe which authors participated in the procedures. It is important to describe how the process was conducted and, if there were disagreements, how they were solved. Validity of the findings: The statement on line 227 is very confusing. What do authors mean by “Of a total of 60 studies, 10 discussed accessibility.”. Considering the topic of the study, how could they not consider accessibility? In fact, further reading the paper reveals that the categories chosen to organize the studies has little justification and explanation in terms of conceptualization. There is very little depth in the description of the categories chosen for the papers. There are several inconsistencies in the citation style that need to be fixed. Unfortunately, the current version of the paper does not present enough consolidation of knowledge in the field, with superficial analyses with little conceptualization to provide a broader overview of the area. In my view, the systematic review would need a new analysis, with a substantially deeper analysis and conceptualization. Additional comments: no comments.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A SYSTEMATIC LITERATURE REVIEW ON THE USABILITY OF MOBILE APPLICATIONS FOR VISUALLY IMPAIRED USERS Review round: 1 Reviewer: 3

Basic reporting: The study aims to find the usability issues for the visually impaired in mobile applications. The authors systematically searched and analyzed the papers on the usability of mobile applications for the visually impaired. The paper reads well. The references used in the Introduction section (Brady et, al and Bastien et. Al.) are pretty old and should be updated. There are no similar literature review papers mentioned in this review. The description and the relevance of the problem should be extended. The authors mention that they classified the studies into five different themes in the Introduction section, but they should also explain how they chose them. Also, there are seven themes in Figure 1. At this point of reading the paper it is not clear what the percentages mean in Figure 1 since this is explained a few pages below. I suggest that this part is excluded from Intruduction section. Experimental design: Was this review registered? I like Table 1, where all research questions are listed and the motivation for them is explained. I don't see the contribution of Figure 2 to better undestandability of the paper and I propose the authors to exclude it. The NAILS project webpage should be cited. »Keyword protocol« in actually not a protocol - it is a search string. Please explain the difference (benefit) in results obtained using the initial and the »new« search string. Table 2 is also not needed. Everything can be explained in one or two sentences. The legend »Series 1« at Figure 4 is not needed. The databases on y-axis should be listed in decreasing order. Figure 9 is not a PRISMA 2009 Flow Diagram. Please refere to the guidances described at http://www.prisma-statement.org/ or. directly at (https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwi0yrv8wqjyAhVImYsKHU5cAGQQFnoECAQQAQ&url=http%3A%2F%2Fprisma-statement.org%2Fdocuments%2FPRISMA%25202009%2520flow%2520diagram.pdf&usg=AOvVaw1qm3ududj_3lshDSZ29LLL) You should mention that you folowed PRISMA 2009 guidances for SR in the begining of the methodology section. Also, mention when did the searcing through the databases took place. Validity of the findings: I think the reader would benefit from the table of selected papers where the authors could maybee present the summarization of the results from Discussion section (or some of them…). I know that 60 papers is a lot (for the SLR), but the reader has to search for information on which papers were selected through the text now. When describing the study which involved participants, please includ information on the sample size (like in line 458; the description of the next study (Rahman et. All) does not include the information on number of participants. This would also be important feature in the table with all selected studies. RV6: Please describe in more details what was the range of sample sizes in the studies and be more specific in reporting the findings with adding the references on the papers. Additional comments: The references should be listed in alphabetic order if the chosen citation style is used.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A SYSTEMATIC LITERATURE REVIEW ON THE USABILITY OF MOBILE APPLICATIONS FOR VISUALLY IMPAIRED USERS Review round: 2 Reviewer: 1

Basic reporting: No comment Experimental design: No comment Validity of the findings: No comment Additional comments: The authors made all the suggested revisions and I am satisfied with their work.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A SYSTEMATIC LITERATURE REVIEW ON THE USABILITY OF MOBILE APPLICATIONS FOR VISUALLY IMPAIRED USERS Review round: 2 Reviewer: 2

Basic reporting: The authors have included almost all of my suggestions in the reviewed manuscript. I still miss a table with a short description of the papers included in the SLR (60 papers). This table could be included in the appendix. Right now I can not even confirm that there were really 60 papers included in the SLR - I would have to extract unique references from the description of the results. Experimental design: The study methodology is adequate and well described. Validity of the findings: The goals are meet and the research questions have been answered and discussed. The future directions are stated. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: TOWARDS AN ARABIC SIGN LANGUAGE (ARSL) CORPUS FOR DEAF DRIVERS Review round: 1 Reviewer: 1

Basic reporting: • English needs to be improved significantly. I suggest a professional proofread, so that your writing can reach and interest a wider audience. Here are some examples: Lines 53-55: “…there are some researchers…provided their effort” Line 139: “These complicities are explained below” Line 286: “Feature Work” Table 2: “أو”. • Considerate language should be used when dealing with inclusiveness issues. Avoid, for example, the word "normal" as opposed to "deaf" (lines 23, 38, 40, 49, 58, 59, 60, 63). • It is better to use acronyms after explaining their meaning. For example: Lines 42-43: (ASL) American Sign Language, (BSL) British Sign Language, (ASL) Australian Sign Language, (ISL) Indian Sign Language and (ArSL) Arabic Sign Language ... could be changed to: American Sign Language (ASL), British Sign Language (BSL), Australian Sign Language (Auslan), Indian Sign Language (ISL), and Arabic Sign Language (ArSL). Note that the correct acronym for Australian Sign Language is Auslan. • Avoid paragraphs that are too long, as it makes reading cumbersome (example: lines 81-118). • The introduction and the background show context. However, lines 92-94 state that "For instance, Support Vector Machine (SVM) with a camera that gained 98.8 93% (Luqman & Mahmoud, 2017) (PCA) Principal Component Analysis, and (HMM) Hidden 94 Markov Model that achieved 99.9% accuracy (Ahmed & Aly, 2014)". Should these percentages be considered baselines? If so, this should be set from the abstract. In fact, the title should be reconsidered, as it indicates that it is an "Implementation", but then it becomes clear that it is a "Proposal". It should be clear from the outset whether the proposal is by any means compared against a baseline. It should be clear whether the proposal is by any means compared against a baseline. If the authors consider that this is not relevant, given that they focus on a new domain (public transport drivers and passengers), they should make it explicit. In the same way, whether it is an implementation or a proposal, it is good to say what it is. Eg: "implementation of a prototype of machine translation" or "proposal of a corpus with linguistic support". • Although it is not a review article, it is good practice to briefly explain how the results presented in the Background have been produced. These search strings can enrich the Background section and it is left to the discretion of the authors to use them: https://scholar.google.com/scholar?hl=es&as_sdt=1%2C5&as_ylo=2017&as_vis=1&q=intitle%3A%22sign+language%22+intitle%3A%22systematic%22&btnG= https://scholar.google.com/scholar?hl=es&as_sdt=1%2C5&as_ylo=2017&as_vis=1&q=intitle%3A%22sign+languages%22+intitle%3A%22systematic%22&btnG= https://scholar.google.com/scholar?hl=es&as_sdt=1%2C5&as_ylo=2017&as_vis=1&q=intitle%3A%22sign+language%22+intitle%3A%22state-of-the-art%22&btnG= • The authorship of the figures must be clear. In some places, the author Abba is quoted and in others Abbas is quoted, causing confusion as to whether this paper is a continuation of earlier research. If so, you should make this clarification from the beginning and explain what is new with respect to the previous work. This is part of the context and it is recommended to place it at the beginning of the abstract. Experimental design: • Methods are described with sufficient detail, in order to replicate. • The journal requests in its guidelines that the research questions be well defined, relevant and meaningful. Please state in the Introduction what the research question is, which could be something like "How can a corpus be built by computer means for ArSL that is valid from a linguistic perspective?". The answer to this question must be made explicit in the Conclusions. Validity of the findings: • The research findings appear to be valid. In any case, it is recommended to clearly indicate in line 279 why a WER of 10.23% is considered good (refer to the related literature). • Underlying data have been provided. However, in "Chapter 8" concerning Amount, it is not clear why the word "Riyals" is omitted from number 5 onwards. Additional comments: • The authors' research is extremely valuable since it addresses a problem of social inclusion in an adequate paradigm, which is that of providing tools for work and incorporation into the productive scheme. Innovation is evident within the public transport domain in Saudi Arabia. Changes are required to achieve the high publishing standards of PeerJ Computer Science, which is why this reviewer has attempted to do a thorough and explicit review.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: TOWARDS AN ARABIC SIGN LANGUAGE (ARSL) CORPUS FOR DEAF DRIVERS Review round: 1 Reviewer: 2

Basic reporting: The introduction is clear and easy to follow, however, there are a few shortcomings. Line 57, “deaf people have some difficulties in communicating with others while they are driving a vehicle like a car and the deaf or normal person sitting as a passenger”. This statement is confusing if this work aims at deaf people driving or deaf people sitting next to a driver? Authors can rephrase this for better understanding to the reader. Literature survey on the ArSL recognition and machine translation is fine, but the novelty of this work is in the corpus. A discussion on existing ArSL corpora like An Arabic Sign Language Corpus for Instructional Language in School (https://eprints.soton.ac.uk/271106/) and their limitations in the deaf driver domain can make the reader understand why this corpus is useful Also, it would be good if authors can provide information related to deaf driver corpus in another sign language if there are any. Fig. 6 didn’t have information on the y-axis. Assuming this as Word Error Rate (WER). Adding axis labels in the charts can ease the readers. Experimental design: What are the problems with current approaches? Why deaf drivers can’t use the existing ArSL for communication during driving? It has been mentioned that some of the words are derived from Saudi sign language and the rest of them are contextual to the domain. How are those words considered contextual to the domain? It has been mentioned that “Some of them are collected from the contextual domain of the normal conversation that done between taxi driver and passengers.”. This process needs to be more elaborate. The corpus should contain all the relevant words in the deaf driver domain without leaving any important information. The corpus has 50% WER in the “Welcome” category but what’s the share of this “welcome” category in the whole corpus. Information about class distribution is important and needs to be included. Validity of the findings: Good to see that corpus files made available for the community. However, annotations can’t be found. Any reason for this? Additional comments: Line 282-285, please check the font style and make it uniform. Add a citation or a URL as a footnote for the VEGAS video editor There are a few grammatical errors in the English language used -For instance: In line 254, “For future work supporting, we added in…”.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: TOWARDS AN ARABIC SIGN LANGUAGE (ARSL) CORPUS FOR DEAF DRIVERS Review round: 2 Reviewer: 1

Basic reporting: I am satisfied with the important changes that the authors have made, in response to my recommendations. Experimental design: I am satisfied with the important changes that the authors have made, in response to my recommendations. Validity of the findings: I am satisfied with the important changes that the authors have made, in response to my recommendations. Additional comments: I am satisfied with the important changes that the authors have made, in response to my recommendations. Clarifications of a cultural nature have also been valuable in order to give the paper a final go-ahead.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: TOWARDS AN ARABIC SIGN LANGUAGE (ARSL) CORPUS FOR DEAF DRIVERS Review round: 2 Reviewer: 2

Basic reporting: Good to see the improvements in Introduction and literature review. The article is now in very good shape and easy to follow. Experimental design: Methods are sufficient to replicate the results. No further changes are required. Validity of the findings: No comment. Additional comments: Interesting research and very useful to the community.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: TOWARDS AN ARABIC SIGN LANGUAGE (ARSL) CORPUS FOR DEAF DRIVERS Review round: 2 Reviewer: 3

Basic reporting: The manuscript need to be checked again with respect to issues on spelling and style. (for ex. Line No 32 - spelling of publish need to be checked again) Uniformity in listing the references is expected. Also it is suggested to cite the latest references in the manuscript. In general, Basic reporting is ok. Experimental design: This research focus on communicating with persons with hearing and speech impairment especially with drivers. The research question could have been presented well. Methodology could have been presented in detail. Validity of the findings: The time complexity of the proposed methodology could have been discussed. Accuracy is not discussed. Conclusion is ok. Additional comments: This research focus on communicating with persons with hearing and speech impairment. what motivates the author to restrict with only drivers. Even the passengers with hearing and speech impairment need to communicate with a driver who is not like that. Title of the manuscript can be checked again with respect to this. Line No 152 need to be checked. deaf people can communicate, but they cannot understand what others say. The time complexity of the proposed methodology could have been discussed. Accuracy could have been discussed. The manuscript need to be checked again with respect to issues on spelling and style. (for ex. Line No 32 - spelling of publish need to be checked again) Uniformity in listing the references and uniformity in citing the references is expected. Some of the articles cited are not listed in the reference. Also few papers in reference are not cited. Also it is suggested to cite the latest references in the manuscript.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SYNTACTIC MODEL-BASED HUMAN BODY 3D RECONSTRUCTION AND EVENT CLASSIFICATION VIA ASSOCIATION BASED FEATURES MINING AND DEEP LEARNING Review round: 1 Reviewer: 1

Basic reporting: In this paper, authors presented a robust approach to human posture analysis and gait event detection from complex video-based data. For this, initially posture information, landmark information are extracted, and human 2D skeleton mesh are extracted, using this information set we reconstruct the human 2D to 3D model. Contextual features, namely, degrees of freedom over detected body parts, joint angle information, periodic and non-periodic motion, and human motion direction flow, are extracted. For features mining, authors applied the rule-based features mining technique and, for gait event detection and classification, the deep learning-based CNN technique is applied over the mpii-video pose, the COCO, and the pose track datasets Experimental design: Experimental Design is Ok. But some more technical revisions are required in this paper with regard to Algorithm and Flowchart of the proposed methodology. Validity of the findings: Data and Analysis experimentation based finiding are Ok. Additional comments: The Paper needs the following Major Revisions: 1. Introduction should be more broad with regard to Problem Definition and even the scope of the paper. More information needs to be discussed in the paper. 2. Related works- In every related works, the stress should be on the novelty proposed, the explanation of the technique and even on experimental conclusions. At the end of related works, highlight in 9-15 lines what overall technical gaps are observed that led to the design of the proposed methodology. 3. Add the Algorithm, Flowchart and Steps of working of the proposed methodology. 4. Add some performance analysis section to this paper. 5. Add future scope to this paper.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SYNTACTIC MODEL-BASED HUMAN BODY 3D RECONSTRUCTION AND EVENT CLASSIFICATION VIA ASSOCIATION BASED FEATURES MINING AND DEEP LEARNING Review round: 1 Reviewer: 2

Basic reporting: Why not compare all methods on three data sets in Table 9? Experimental design: no comment Validity of the findings: no comment Additional comments: 1 According to the performance scores given in the tables, the results of the proposed method are compelling. But from a technical perspective, it is mainly a combination of existing technologies, and the contributions are not obvious. I hope that the manuscript will highlight the contributions of this paper in the revised version, omitting unnecessary formulas. 2 Please discuss the failure cases and the limitations of the proposed method. 3 Why not compare all methods on three data sets in Table 9? 4 “To find the knees points we take the midpoint between the human-body midpoint and the two feet points. Eq. (6) demonstrates the human knee points;” According to this manuscript at lines 204-206, the denominator in formula (1) is 2? 5 My personal suggestion is to intersperse the Figures in the manuscript. 6 Some paragraphs are indented too much (such as the first paragraph), and some are not indented.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SYNTACTIC MODEL-BASED HUMAN BODY 3D RECONSTRUCTION AND EVENT CLASSIFICATION VIA ASSOCIATION BASED FEATURES MINING AND DEEP LEARNING Review round: 1 Reviewer: 3

Basic reporting: 1. The writing in this article lacks proper structure, has various grammatical mistakes, and somethings are not clearly formulated. 2. The mathematical notation used in the manuscript is not easy to parse and doesn't follow standard notations, which makes it difficult for users to understand the ideas in detail. This also results in various redundancies throughout the manuscript. I have marked some of them in the paper and attached them. Authors should make the manuscript easy for readers to follow. 3. Limitations and future works are not discussed. 4. All the figures and tables are added towards the end of the manuscript and are not placed correctly. Captions for the figures are not self-explanatory. 5. Some variables are not defined before they are used (see the attachment). 6. Relevant missing work: Openpose, Cao et. al. 7. Feature generation steps have various ambiguities and require clarification(see the attachment). 8. Algorithms should be re-written and some of them(like Algorithm 2) are not even required. Indention is not followed in algorithms Experimental design: 1. Goal and idea of the method is interesting. Authors propose various features like pose, joint angle, silhouette, etc, and use these for gait-event classification. Although all the features used seem important for gait classification, there can be some redundant features and some might enable wrong correlation. I would be very interested to see a comprehensive analysis of feature selection. 2. Since the method section is not clear, even though the techniques used in the method are trivial methods, I fear, paper cannot be exactly reproduced. This is because some things were not clear. 3. I have marked specific questions in the manuscript. Validity of the findings: 1. Datasets used in the paper are publicly available datasets and authors provide detailed evaluation for various types of motion. 2. Limitations and future works should be discussed in brief. Authors do not explain the limitations of their method like multi-step data preprocessing step, time etc. 3. Authors use CNN for classification, although CNN is powerful features extractors, how do author justify their choice of not using CNN based features? A comparative study or justification is required. 4. How is this pipeline of even gait classification better than OpenPose based 2d joint detection + gait classification layer. This is the very basic experimental baseline that comes to my mind, given OpenPose performs very well for most of the datasets and generalizes well. Additional comments: I have corrected some grammatical mistakes and anomalies, not all. So you should carefully read the manuscript again along the same line.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SYNTACTIC MODEL-BASED HUMAN BODY 3D RECONSTRUCTION AND EVENT CLASSIFICATION VIA ASSOCIATION BASED FEATURES MINING AND DEEP LEARNING Review round: 2 Reviewer: 1

Basic reporting: In this paper, authors presented a robust approach to human posture analysis and gait event detection from complex video-based data. For this, initially posture information, landmark information are extracted, and human 2D skeleton mesh are extracted, using this information set we reconstruct the human 2D to 3D model. Contextual features, namely, degrees of freedom over detected body parts, joint angle information, periodic and non-periodic motion, and human motion direction flow, are extracted. For features mining, authors applied the rule-based features mining technique and, for gait event detection and classification, the deep learning-based CNN technique is applied over the mpii-video pose, the COCO, and the pose track datasets. Experimental design: Experimental design is Ok and falls under aim and scope of the paper. Validity of the findings: Yes, the revised findings are Ok. Additional comments: The Paper stands Accepted with no further rrevisions.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SYNTACTIC MODEL-BASED HUMAN BODY 3D RECONSTRUCTION AND EVENT CLASSIFICATION VIA ASSOCIATION BASED FEATURES MINING AND DEEP LEARNING Review round: 2 Reviewer: 2

Basic reporting: Following the suggestions and requests of all reviewers, the authors have updated the manuscript. The problem statement, technical gaps, experiments, algorithm, and flow charts have been updated/added. This improves the initial submission by a great margin. The paper reads well now. Experimental design: no comment Validity of the findings: My previous concern of using CNN not just for classification but also for features has not been answered well. Also, authors do now validate their method against the openpose+CNN classifier baseline. They only provide some argument, which doesn't seem strong enough. I would recommend the authors add openpose+CNN classifier baseline results in the paper, even if their method performs worse than this. It will be useful for readers to draw a conclusion. Additional comments: My only request is that the authors add the above baseline(section 3) in the paper.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A FAST VECTORIZED SORTING IMPLEMENTATION BASED ON THE ARM SCALABLE VECTOR EXTENSION (SVE) Review round: 1 Reviewer: 1

Basic reporting: This paper proposes a fast vectorized sorting implementation based on the ARM scalable vector extension (SVE). It ports the author's previous work about the AVX-SORT algorithm to SVE, defines a new Bitonic-sort variant using SVE, and implements an efficient Quicksort variant using OpenMP tasks. Strengths: The description is clear. The work seems technically sound and leads to impressive experimental results with raw data shared. Weaknesses: (1) It is good the author declares this work is to port the previous work in AVX to SVE and cites the paper "A Novel Hybrid Quicksort Algorithm Vectorized using AVX-512 on Intel Skylake". There is no cover letter so that the reviewer is not certain whether it is allowable to directly reuse figures 1, 2 and 3. Besides, it is not clear the difference between this paper and the reference paper. (2) This paper is mainly about SVE, so it is better to not only mention AVX in the related work. (3) The performance figures, especially figure 4 and figure 7, are very hard to read; e.g., there are two markers that are too similar in figure 4, and lines with the same color but different width in figure 7. (4) As declared in the paper, there are three contributions. It would be good if the author could detail the last one, "Implement an efficient Quicksort variant using OpenMP tasks", because there are only very brief descriptions about this in the paper. (5) What's the difference between SVE-Bitonic and SVE512-Bitonic? Is SVE512-Bitonic a version of SVE-Bitonic with fixed vector length? If so, the performance is a little confusing as the performance of SVE512-Bitonic will not be worse than SVE-Bitonic, right? (6) In figure 7, there are several lines of different numbers of threads are overlapped, which means the overheads and benefits of different numbers of threads on the same problem are the same. The reviewer was wondering why this happens. (7) It would be a plus if the author could define how good the improvement is by using SVE in this paper, e.g., comparing it to the theoretical one instead of only to AVX. Experimental design: no comment Validity of the findings: no comment Additional comments: no comment

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A FAST VECTORIZED SORTING IMPLEMENTATION BASED ON THE ARM SCALABLE VECTOR EXTENSION (SVE) Review round: 1 Reviewer: 2

Basic reporting: 1. This paper need to add more information about Scalable Vector Extension in introduction and related part. Especially in section 2.2 and 2.3. It feels like the author is writing a paper about AVX instead of SVE. In section 2.2.1 why use avx512 as an example instead of SVE? SVE related work and publications are not mentioned in section 2.3. 2. Figure 4 is very confusing and hard to read. Choose different lines and markers will be better. Experimental design: 1. In line 365 the author claims that "This means that considering a static vector size of 512 bits, with compare-exchange indices hard coded and no loops/branches, does not provide any benefit, and is even slower for more than 70 values." The author needs to provide more information and details to support this assumption, because normally extra branching instruction may need extra cycles. It will be helpful if the author can provide low level explanation, such as assemble code snippets. Also some performance tools 2. The vector length of A64FX is configurable, I would like to see the results of the same experiments by setting vector length to 256 bits, then we can evaluate performance benefits with different vector length. Validity of the findings: "We also demonstrate that an implementation designed for a fixed size of vectors is less efficient", I would like to see more evidence to support this conclusion. Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: A FAST VECTORIZED SORTING IMPLEMENTATION BASED ON THE ARM SCALABLE VECTOR EXTENSION (SVE) Review round: 2 Reviewer: 1

Basic reporting: Thanks for the authors' detailed response, and there is one remaining question. In the description of Figure 7, it's said "using more than 8 threads does not provide any benefit", which is not true when the number of values is large, as shown in the figure. Minor: (1) Figure 7 is split so that maybe markers could be used to differentiate lines instead of line width. (2) there is only one subsubsection (2.2.1), and maybe it's better to remove it. Experimental design: no comment Validity of the findings: no comment Additional comments: No additional comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: NFAD: FIXING ANOMALY DETECTION USING NORMALIZING FLOWS Review round: 1 Reviewer: 1

Basic reporting: The introduction could be more detailed, such as adding the reason why the proposed method in the manuscript works. I suggest that the authors add more recent and closely related references on the same problem, such as: (1) Feature Encoding with AutoEncoders for Weakly-supervised Anomaly Detection, IEEE Transactions on Neural Networks and Learning Systems, 2021. (2) Deep Weakly-supervised Anomaly Detection. arXiv preprint: 1910.13601 (2019). I also encourage the authors to consider comparing their performance with those of the two citations I recommended. The authors had use an early version of the second recommended citation as a comparing method(devnet); the first citation has a superior performance than that of devnet on the same problem. The first paper is my own paper. And the second one has no conflict of interest with me. The manuscript is not well formated in the current version. Also, there are some typos and errors in the content, e.g., the caption of some tables are wrongly represented as "Figure X.". Excludingly, this manuscript basically meets the requirements. Experimental design: In the experiments, the analysis of influencing factors is missing, and the author can add some experiments to this regard. Validity of the findings: More experiments are suggested to conduct to show the effect of the proposed method as well as the major factors that may have impact on its performance. Additional comments: I attempted to run the code, but an error occured. Maybe there's something wrong with my environment or my network. But I hope the authors make sure the code could run automatically without any additional change. Also for this code, the authors are suggested to provide more detailed instructions.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: NFAD: FIXING ANOMALY DETECTION USING NORMALIZING FLOWS Review round: 1 Reviewer: 2

Basic reporting: Clear and unambiguous, professional English used throughout. The language is correct and the article is written in clear and professional English. Literature references, sufficient field background/context provided. The literature on related papers could be more explored. Just a few numbers of papers were cited, and the article mentions just a few methods of AD, specially based on classification, while there are others as ones based on Clustering, distance and density (as in F. Meng, G. Yuan, S. Lv, Z. Wang, and S. Xia, “An overview on trajectory outlier detection,” Artificial Intelligence Review, vol. 52, no. 4, pp. 2437–2456, 2019). Professional article structure, figures, tables. Raw data shared. The structure doesn’t follow the format suggest by the journal. It combines “Results” and “Discussion” in just one item, what makes the paper less clear, because there are discussions presented before results. Result are present in 3 paragraphs after Figure 4 and before figures 7 to 9. In Figure 1, specially the last diagram is too hard to see the tail p-values. Figures 4, 5, 6, 7, 8 and 9 are not figures but table. They should shave been mentioned as tables not figures. Figures 8 and 9 are misplaced in the text. They are in “Conclusion”, but they should be included in “Results and Discussions” as figures 4 to 7 are. In figure 7 and 8, the best results are bolded as in other figures. They should have been bolded despite the best results are not provided by the proposed algorithm. All appropriate raw data have been made available in accordance with our Data Sharing policy. Self-contained with relevant results to hypotheses. The article proposes a new technique to address DA problem. The problems are well stated, and the hypothesis is tested, showing relevant results, even the ones that the proposed algorithm doesn’t performance well. The submission is ‘self-contained,’ and represents an appropriate ‘unit of publication’. It also includes all results relevant to the hypothesis. Formal results should include clear definitions of all terms and theorems, and detailed proofs. In topic 3, when authors state that normalizing flows generative mode aims to fit the exact probability distribution of data, they don’t mention that normalizing flows provides exact inference and log-likelihood evaluation as its merits (D. P. Kingma and P. Dhariwal, “Glow: Generative flow with invertible 1x1 convolutions,” in Advances in Neural Information Processing Systems, 2018, pp. 10 215–10 224). It is important because when the model is trained in topic 4, it is trained by standard for normalizing scheme of maximization the log-likelihood, which is not described previously when normalizing flows are described (Topic 3). Experimental design: Original primary research within Aims and Scope of the journal. The submission is within Aims and Scope of the journal, because proposes a new algorithm to AD task, a field related to Computer Sciences. The paper is a Research Article, in which a well stated hypothesis is tested and the results of these tests are presented. Research question well defined, relevant & meaningful. It is stated how research fills an identified knowledge gap. The research question is well defined and also the article scope. However, in the second line of “Problem Statement” item, the statement “In practice, while such samples are, indeed, most likely to be anomalous, often some anomalies might not be distinguishable from normal samples” doesn’t have any reference. Also, the effect of the number of anomaly samples in the training data should be addressed more deeply, as in Exploring normalizing flows for Anomaly Detection (Pathak, C. 2019), as this effect is one of the most important issues in the paper. Rigorous investigation performed to a high technical & ethical standard. The investigation must have been conducted rigorously and to a high technical standard. The research must have been conducted in conformity with the prevailing ethical standards in the field. Methods described with sufficient detail & information to replicate. Methods should be described with sufficient information to be reproducible by another investigator. Validity of the findings: Impact and novelty not assessed. Meaningful replication encouraged where rationale & benefit to literature is clearly stated. Conclusions are based in a metric presented in tables, so we can see clearly where the proposed algorithm surpass others and where it doesn’t. However, the metric employed is not clearly described. We infer that authors use accuracy, but this option is not justified. All underlying data have been provided; they are robust, statistically sound, & controlled. The data is provided and it was possible to rerun the experiment. Conclusions are well stated, linked to original research question & limited to supporting results. The conclusions are appropriately stated, and connected to the original question investigated, and are supported by the results, as based in obtained results described in figures 4 to 9. The proposed algorithm outperformances many tested algorithms in most of situations, except when dialing with images dataset. Additional comments: As general conclusion, the paper can be published after mandatory corrections, specially these described in item 1.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: NFAD: FIXING ANOMALY DETECTION USING NORMALIZING FLOWS Review round: 2 Reviewer: 1

Basic reporting: It seems that the authors have addressed most my concerns. I suggest to a minor revision for the manuscript in its current form. The comments are as follows:1 1. for the two references, "(1) Feature Encoding with AutoEncoders for Weakly-supervised Anomaly Detection, IEEE Transactions on Neural Networks and Learning Systems, 2021. (2) Deep Weakly-supervised Anomaly Detection. arXiv preprint: 1910.13601 (2019)." What is the major technical differences between the references and the proposed method? I guess that the authors did not display the bad cases for the proposed method, which the authors should present and highlight them. 2. some references are not complete, e.g., ““Zhou, Y., Song, X., Zhang, Y., Liu, F., Zhu, C., and Liu, L. (2021). Feature encoding with autoencoders for weakly-supervised anomaly detection.””, which lacks the publication title and etc. "241 Chen, R. T. Q., Behrmann, J., Duvenaud, D., and Jacobsen, J.-H. (2019). Residual flows for invertible 242 generative modeling." 3 a detailed lists of revisions are needed in the response letter for each coments in the previous review, so that the reviewer can see the revisions more clear. Experimental design: please see the above Validity of the findings: please see the above Additional comments: no

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EVENT CLASSIFICATION FROM THE URDU LANGUAGE TEXT ON SOCIAL MEDIA Review round: 1 Reviewer: 1

Basic reporting: The title of the paper is “Event classification from text existing on social media” it's not mentioned in the title and abstract, that the data is in Urdu language !!! Experimental design: In the lines 182-183 the authors mentioned that “ In comparison, we used sentences of Urdu language for classification and explored the textual features of sentences. We disclosed all textual features in detail in this paper that were not reported ever in state-of-art.” Comment : The authors didn’t mention in details all the textual features for the Urdu language, and they didn't provide enough examples for each feature so the reader can understand it. the authors didn't add a section for the machine learning algorithms they used in the paper, or even add a references for these algorithms!!!! the authors didn't mentioned the stemmer or POS tool they used in the experiments. the authors mentioned "The dataset consists of more than 1 million (1,02,962) labeled instances of twelve (12) 22 different types of events." what was the size of training data , and the size for testing data!!!! the authors didn't include any example for their work!!! Validity of the findings: the authors didn't experiment all choices of the extracted features, and they just did the experiment on "title and last 4 word" !!! only the tables 7-10 not explained. All the figures not explained. the findings not clear and accurate!!! Additional comments: NO comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EVENT CLASSIFICATION FROM THE URDU LANGUAGE TEXT ON SOCIAL MEDIA Review round: 1 Reviewer: 2

Basic reporting: The author works in the text classification domain and performs an experiment on the event classification in the Urdu language. The author used state of the art machine learning models and show that the KNN outperform all machine learning models. The author used a dataset contain 1 million records and 12 different labels taken from social me dia. Experimental design: 1-The author used TF-IDF and count vectorizer for features extraction. Why author used only TF-IDF and BoW count vectorizer? the author should use word2vec in comparison with TF-IDF and count vectorizer see the article "Rustam, F., Ashraf, I., Mehmood, A., Ullah, S. and Choi, G.S., 2019. Tweets classification on the basis of sentiments for US airline companies. Entropy, 21(11), p.1078." 2- Author should check the IDF formula again. 3- KNN is best performer when features set will be small. Find a supporting article for that where knn outperform on a large features set. 4- Author used state of the art method which have already lots of used in classification what is the novality of author work. work in specific domain is not a novality. Validity of the findings: No comment Additional comments: 1- The experimental diagram should follow the full experimental flow. 2- author work on specific language how to deal with preprocessing of text there should be a more clear discussion about the library and self-generated library if anything used by the author. 3- the author should perform a comparison with results without preprocessing of text. 4- grammar should be checked thoroughly.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EVENT CLASSIFICATION FROM THE URDU LANGUAGE TEXT ON SOCIAL MEDIA Review round: 2 Reviewer: 1

Basic reporting: no comment Experimental design: no comment Validity of the findings: The accuracy for the ML algorithms with the features , are not reasonable , increasing from 17% to 85% , then the final results 99% !!! The authors didn't mention any weaknesses of their model. In many English and Arabic researches , the SVM and KNN , were the best ML algorithms , but in your research they are 's not, can the authors mention the reasons!! Additional comments: The examples should be written in Urdu , and English language as well,

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EVENT CLASSIFICATION FROM THE URDU LANGUAGE TEXT ON SOCIAL MEDIA Review round: 2 Reviewer: 2

Basic reporting: Comments are below Experimental design: Comments are below Validity of the findings: Comments are below Additional comments: The author asks to revise the manuscript to improve the quality of the article and for this, I have mentioned some comments for the author. Lots of comments have cooperated but still, there are some weak areas in the manuscript. 1- Abstract's first 3 and 4 lines are too general and the overall abstract should be more attractive. best performers' results should be added to the abstract. 2- Results section should contain more detail about the results and represent the significance of the models.

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EVENT CLASSIFICATION FROM THE URDU LANGUAGE TEXT ON SOCIAL MEDIA Review round: 4 Reviewer: 1

Basic reporting: No comments Experimental design: No comments Validity of the findings: no comments Additional comments: no comments

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: EVENT CLASSIFICATION FROM THE URDU LANGUAGE TEXT ON SOCIAL MEDIA Review round: 4 Reviewer: 2

Basic reporting: The author presented the event classification for the Urdu language text existing on social media and news channels. The dataset contains more than 0.1 million (102,962) labeled instances of twelve (12) different types of events. Title, Length, and last-4-words of a sentence are used as features to classify events. The Term Frequency-Inverse Document Frequency (tf-idf) showed the best results as a feature vector to evaluate the performance of the six popular machine learning classifiers. The author resolves the comments but still lots of things should be improved especially the abstract. such as: These mention two sentences are the same in the abstract why? Random Forest (RF), Decision Tree, and k-Nearest Neighbor outperformed among the other classifiers. Random Forest and K-Nearest Neighbor are the classifiers that out-performed among other classifiers by achieving 98.00% and 99.00% accuracy, respectively. The Abstract didn't contain detail about contribution and methodology. it's too general and short. Experimental design: No commets Validity of the findings: The dataset is imbalanced then the results are too significant justifies that the models are no overfitted on majority class data. The author gives the confusion matrix report in terms of TP, TN, FP, and FP.. These terms are useful for binary classification how they adjust them in multiclass classification. add some visual infromation. Additional comments: No comemnt

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SMALL FACIAL IMAGE DATASET AUGMENTATION USING CONDITIONAL GANS BASED ON INCOMPLETE EDGE FEATURE INPUT Review round: 1 Reviewer: 1

Basic reporting: In this paper, a new image-to-image translation framework using conditional GANs is proposed, which can generate diverse photorealistic images from limited edge features after training with a small number of training images.The article has a certain degree of innovation and novelty, the experiment is relatively sufficient, and the overall content is complete, but there are still some problems that the author hopes to correct： 1）Whether the pictures involved in the article can be directly reflected in the text to facilitate reading and understanding; 2）The quality of the picture is not very high, it is recommended to replace it, and some of the picture content is redundant, and it is recommended to delete it; 3）Figure 16 compares the KID and FID values of different methods within 500 sheets, and the difference between the results of the other two methods and the method in this article is getting smaller and smaller. Can the author compare more images, or briefly Explain the change trend of subsequent FID and KID results; Experimental design: no comment Validity of the findings: no comment Additional comments: no comment

You are one of the reviewers, your task is to write a review for the article. You will be given the title of the article, the number of the round in which the article is located, and your order among the reviewers.

Title: SMALL FACIAL IMAGE DATASET AUGMENTATION USING CONDITIONAL GANS BASED ON INCOMPLETE EDGE FEATURE INPUT Review round: 1 Reviewer: 2

Basic reporting: It is better to include a paragraph providing details about the structure of the paper (could be the last paragraph of the Introduction section). The related work section is very comprehensive (good work!). Include some details of additional image data augmentation works would further improve the related work section. Indicate potential future works of this research in the conclusion section. Experimental design: Structuring the methods section is very important to enhance the readability of the paper. It is better to include individual subsections to three major parts of the proposed method. Lack of supportive arguments is the primary concern in the results section. Authors will get benefitted by including a comprehensive discussion on both qualitative and quantitative results. An ablative study is a must in this case to express the significance of each module. Validity of the findings: Highlight the limitations of the work is important for researchers in the field to explore more in this domain of interest. Discuss some of them in the results section. Additional comments: No additional comments