Datasets:

feradauto

/

NLP4SGPapers

like 2

sido-etal-2021-czert

https://aclanthology.org/2021.ranlp-1.149

Czert -- Czech BERT-like Model for Language Representation

This paper describes the training process of the first Czech monolingual language representation models based on BERT and ALBERT architectures. We pre-train our models on more than 340K of sentences, which is 50 times more than multilingual models that include Czech data. We outperform the multilingual models on 9 out of 11 datasets. In addition, we establish the new state-of-the-art results on nine datasets. At the end, we discuss properties of monolingual and multilingual models based upon our results. We publish all the pretrained and fine-tuned models freely for the research community.

This work has been partly supported by ERDF "Research and Development of Intelligent Components of Advanced Technologies for the Pilsen Metropolitan Area (InteCom)" (no.: CZ.02.1.01/0.0/0.0/17 048/0007267); and by Grant No. SGS-2019-018 Processing of heterogeneous data and its specialized applications. Computational resources were supplied by the project "e-Infrastruktura CZ" (e-INFRA LM2018140) provided within the program Projects of Large Research, Development and Innovations Infrastructures.

2021

gehrmann-etal-2021-gem

https://aclanthology.org/2021.gem-1.10

The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics

We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly evolving ecosystem of automated metrics, datasets, and human evaluation standards. Due to this moving target, new models often still evaluate on divergent anglo-centric corpora with wellestablished, but flawed, metrics. This disconnect makes it challenging to identify the limitations of current models and opportunities for progress. Addressing this limitation, GEM provides an environment in which models can easily be applied to a wide set of tasks and in which evaluation strategies can be tested. Regular updates to the benchmark will help NLG research become more multilingual and evolve the challenge alongside models. This paper serves as the description of the data for which we are organizing a shared task at our ACL 2021 Workshop and to which we invite the entire NLG community to participate.

The authors of this paper not named in the groups participated in initial discussions, participated in the surveys, and provided regular feedback and guidance. Many participants commented on and helped write this paper. We additionally thank all participants of INLG 2019, the Generation Birdsof-a-Feather meeting at ACL 2020, the EvalNL-GEval Workshop at INLG 2020, and members of the generation challenge mailing list of SIGGEN for their participation in the discussions that inspired and influenced the creation of GEM.

2021

novak-novak-2021-transfer

https://aclanthology.org/2021.ranlp-1.119

Transfer-based Enrichment of a Hungarian Named Entity Dataset

In this paper, we present a major update to the first Hungarian named entity dataset, the Szeged NER corpus. We used zero-shot crosslingual transfer to initialize the enrichment of entity types annotated in the corpus using three neural NER models: two of them based on the English OntoNotes corpus and one based on the Czech Named Entity Corpus fine-tuned from multilingual neural language models. The output of the models was automatically merged with the original NER annotation, and automatically and manually corrected and further enriched with additional annotation, like qualifiers for various entity types. We present the evaluation of the zero-shot performance of the two OntoNotes-based models and a transformer-based new NER model trained on the training part of the final corpus. We release the corpus and the trained model.

This research was implemented with support provided by grants FK 125217 and PD 125216 of the National Research, Development and Innovation Office of Hungary financed under the FK 17 and PD 17 funding schemes as well as through the Artificial Intelligence National Excellence Program (grant no.: 2018-1.2.1-NKP-2018-00008).

2021

hu-etal-2019-texar

https://aclanthology.org/P19-3027

Texar: A Modularized, Versatile, and Extensible Toolkit for Text Generation

We introduce Texar, an open-source toolkit aiming to support the broad set of text generation tasks that transform any inputs into natural language, such as machine translation, summarization, dialog, content manipulation, and so forth. With the design goals of modularity, versatility, and extensibility in mind, Texar extracts common patterns underlying the diverse tasks and methodologies, creates a library of highly reusable modules and functionalities, and allows arbitrary model architectures and algorithmic paradigms. In Texar, model architecture, inference, and learning processes are properly decomposed. Modules at a high concept level can be freely assembled or plugged in/swapped out. Texar is thus particularly suitable for researchers and practitioners to do fast prototyping and experimentation. The versatile toolkit also fosters technique sharing across different text generation tasks. Texar supports both TensorFlow and PyTorch, and is released under Apache License 2.0 at https: //www.texar.io. 1

2019

zhao-huang-1998-quasi

https://aclanthology.org/C98-1001

A Quasi-Dependency Model for Structural Analysis it of Chinese BaseNPs

The paper puts forward a quasidependency model for structural analysis of Chinese baseNPs and a MDL-based algorithm for quasidependency-strength acquisition. The experiments show that the proposed model is more suitable for Chinese baseNP analysis and the proposed MDLbased algorithm is superior to the traditional MLbased algorithm. The paper also discusses the problem of incorporating the linguistic knowledge into the above statistical model.

1998

filimonov-harper-2007-recovery

https://aclanthology.org/D07-1065

Recovery of Empty Nodes in Parse Structures

In this paper, we describe a new algorithm for recovering WH-trace empty nodes. Our approach combines a set of handwritten patterns together with a probabilistic model. Because the patterns heavily utilize regular expressions, the pertinent tree structures are covered using a limited number of patterns. The probabilistic model is essentially a probabilistic context-free grammar (PCFG) approach with the patterns acting as the terminals in production rules. We evaluate the algorithm's performance on gold trees and parser output using three different metrics. Our method compares favorably with state-of-the-art algorithms that recover WH-traces.

We would like to thank Ryan Gabbard for providing us output from his algorithm for evaluation. We would also like to thank the anonymous reviewers for invaluable comments. This material is based upon work supported by the Defense Advanced Research Projects Agency (DARPA) under Contract No. HR0011-06-C-0023. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of DARPA.

2007

linzen-2020-accelerate

https://aclanthology.org/2020.acl-main.465

How Can We Accelerate Progress Towards Human-like Linguistic Generalization?

This position paper describes and critiques the Pretraining-Agnostic Identically Distributed (PAID) evaluation paradigm, which has become a central tool for measuring progress in natural language understanding. This paradigm consists of three stages: (1) pretraining of a word prediction model on a corpus of arbitrary size; (2) fine-tuning (transfer learning) on a training set representing a classification task; (3) evaluation on a test set drawn from the same distribution as that training set. This paradigm favors simple, low-bias architectures, which, first, can be scaled to process vast amounts of data, and second, can capture the fine-grained statistical properties of a particular data set, regardless of whether those properties are likely to generalize to examples of the task outside the data set. This contrasts with humans, who learn language from several orders of magnitude less data than the systems favored by this evaluation paradigm, and generalize to new tasks in a consistent way. We advocate for supplementing or replacing PAID with paradigms that reward architectures that generalize as quickly and robustly as humans.

2020

jin-hauptmann-2002-new

https://aclanthology.org/C02-1137

A New Probabilistic Model for Title Generation

Title generation is a complex task involving both natural language understanding and natural language synthesis. In this paper, we propose a new probabilistic model for title generation. Different from the previous statistical models for title generation, which treat title generation as a generation process that converts the 'document representation' of information directly into a 'title representation' of the same information, this model introduces a hidden state called 'information source' and divides title generation into two steps, namely the step of distilling the 'information source' from the observation of a document and the step of generating a title from the estimated 'information source'. In our experiment, the new probabilistic model outperforms the previous model for title generation in terms of both automatic evaluations and human judgments.

The authors are grateful to the anonymous reviewers for their comments, which have helped improve the quality of the paper. This material is based in part on work supported by National Science Foundation under Cooperative Agreement No. IRI-9817496. Partial support for this work was provided by the National Science Foundation's National Science, Mathematics, Engineering, and Technology Education Digital Library Program under grant DUE-0085834. This work was also supported in part by the Advanced Research and Development Activity (ARDA) under contract number MDA908-00-C-0037. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation or ARDA.

2002

kozhevnikov-titov-2014-cross

https://aclanthology.org/P14-2095

Cross-lingual Model Transfer Using Feature Representation Projection

We propose a novel approach to crosslingual model transfer based on feature representation projection. First, a compact feature representation relevant for the task in question is constructed for either language independently and then the mapping between the two representations is determined using parallel data. The target instance can then be mapped into the source-side feature representation using the derived mapping and handled directly by the source-side model. This approach displays competitive performance on model transfer for semantic role labeling when compared to direct model transfer and annotation projection and suggests interesting directions for further research.

The authors would like to acknowledge the support of MMCI Cluster of Excellence and Saarbrücken Graduate School of Computer Science and thank the anonymous reviewers for their suggestions.

2014

knight-sproat-2009-writing

https://aclanthology.org/N09-4008

Writing Systems, Transliteration and Decipherment

Kevin Knight (USC/ISI) Richard Sproat (CSLU/OHSU) Nearly all of the core data that computational linguists deal with is in the form of text, which is to say that it consists of language data written (usually) in the standard writing system for the language in question. Yet surprisingly little is generally understood about how writing systems work. This tutorial will be divided into three parts. In the first part we discuss the history of writing and introduce a wide variety of writing systems, explaining their structure and how they encode language. We end this section with a brief review of how some of the properties of writing systems are handled in modern encoding systems, such as Unicode, and some of the continued pitfalls that can occur despite the best intentions of standardization. The second section of the tutorial will focus on the problem of transcription between scripts (often termed "transliteration"), and how this problem-which is important both for machine translation and named entity recognition-has been addressed. The third section is more theoretical and, at the same time we hope, more fun. We will discuss the problem of decipherment and how computational methods might be brought to bear on the problem of unlocking the mysteries of as yet undeciphered ancient scripts. We start with a brief review of three famous cases of decipherment. We then discuss how techniques that have been used in speech recognition and machine translation might be applied to the problem of decipherment. We end with a survey of the as-yet undeciphered ancient scripts and give some sense of the prospects of deciphering them given currently available data.

2009

holan-etal-1998-two

https://aclanthology.org/W98-0503

Two Useful Measures of Word Order Complexity

This paper presents a class of dependency-based forreal grammars (FODG) which can be parametrized by two different but similar measures of nonprojectivity. The measures allow to formulate constraints on the degree of word-order freedom in a language described by a FODG. We discuss the problem of the degree of word-order freedom which should be allowed ~, a FODG describing the (surface) syntax of Czech.

1998

mohammad-etal-2018-semeval

https://aclanthology.org/S18-1001

SemEval-2018 Task 1: Affect in Tweets

We present the SemEval-2018 Task 1: Affect in Tweets, which includes an array of subtasks on inferring the affectual state of a person from their tweet. For each task, we created labeled data from English, Arabic, and Spanish tweets. The individual tasks are: 1. emotion intensity regression, 2. emotion intensity ordinal classification, 3. valence (sentiment) regression, 4. valence ordinal classification, and 5. emotion classification. Seventy-five teams (about 200 team members) participated in the shared task. We summarize the methods, resources, and tools used by the participating teams, with a focus on the techniques and resources that are particularly useful. We also analyze systems for consistent bias towards a particular race or gender. The data is made freely available to further improve our understanding of how people convey emotions through language.

2018

miculicich-henderson-2022-graph

https://aclanthology.org/2022.findings-acl.215

Graph Refinement for Coreference Resolution

The state-of-the-art models for coreference resolution are based on independent mention pairwise decisions. We propose a modelling approach that learns coreference at the documentlevel and takes global decisions. For this purpose, we model coreference links in a graph structure where the nodes are tokens in the text, and the edges represent the relationship between them. Our model predicts the graph in a non-autoregressive manner, then iteratively refines it based on previous predictions, allowing global dependencies between decisions. The experimental results show improvements over various baselines, reinforcing the hypothesis that document-level information improves conference resolution.

This work was supported in part by the Swiss National Science Foundation, under grants 200021_178862 and CRSII5_180320.

2022

prasad-etal-2008-towards

https://aclanthology.org/I08-7010

Towards an Annotated Corpus of Discourse Relations in Hindi

We describe our initial efforts towards developing a large-scale corpus of Hindi texts annotated with discourse relations. Adopting the lexically grounded approach of the Penn Discourse Treebank (PDTB), we present a preliminary analysis of discourse connectives in a small corpus. We describe how discourse connectives are represented in the sentence-level dependency annotation in Hindi, and discuss how the discourse annotation can enrich this level for research and applications. The ultimate goal of our work is to build a Hindi Discourse Relation Bank along the lines of the PDTB. Our work will also contribute to the cross-linguistic understanding of discourse connectives.

2008

kunz-etal-2021-heicic

https://aclanthology.org/2021.motra-1.2

HeiCiC: A simultaneous interpreting corpus combining product and pre-process data

This paper presents HeiCIC, a simultaneous interpreting corpus that comprises audio files, time-aligned transcripts and corresponding preparation material complemented by annotation layers. The corpus serves the pursuit of a range of research questions focusing on strategic cognitive load management and its effects on the interpreting output. One research objective is the analysis of semantic transfer as a function of problem triggers in the source text which represent potential cognitive load peaks. Another research approach correlates problem triggers with solution cues in the visual support material used by interpreters in the booth. Interpreting strategies based on this priming reduce cognitive load during SI.

2021

chang-etal-2021-nao

https://aclanthology.org/2021.ccl-1.57

脑卒中疾病电子病历实体及实体关系标注语料库构建(Corpus Construction for Named-Entity and Entity Relations for Electronic Medical Records of Stroke Disease)

This paper discussed the labeling of Named-Entity and Entity Relations in Chinese electronic medical records of stroke disease, and proposes a system and norms for labeling entity and entity relations that are suitable for content and characteristics of electronic medical records of stroke disease. Based on the guidance of the labeling system and norms, this carried out several rounds of manual tagging and proofreading and completed the labeling of entities and relationships more than 1.5 million words. The entity and entity relationship tagging corpus of stroke electronic medical record(Stroke Electronic Medical Record entity and entity related Corpus, SEMRC)is fromed. The constructed corpus contains 10,594 named entities and 14,457 entity relationships. The consistency of named entity reached 85.16%, and that of entity relationship reached 94.16%.

Good Health and Well-Being

2021

lee-chang-2003-acquisition

https://aclanthology.org/W03-0317

Acquisition of English-Chinese Transliterated Word Pairs from Parallel-Aligned Texts using a Statistical Machine Transliteration Model

This paper presents a framework for extracting English and Chinese transliterated word pairs from parallel texts. The approach is based on the statistical machine transliteration model to exploit the phonetic similarities between English words and corresponding Chinese transliterations. For a given proper noun in English, the proposed method extracts the corresponding transliterated word from the aligned text in Chinese. Under the proposed approach, the parameters of the model are automatically learned from a bilingual proper name list. Experimental results show that the average rates of word and character precision are 86.0% and 94.4%, respectively. The rates can be further improved with the addition of simple linguistic processing.

2003

zilio-etal-2017-using

https://doi.org/10.26615/978-954-452-049-6_107

Using NLP for Enhancing Second Language Acquisition

Quality Education

2017

nasr-rambow-2004-simple

https://aclanthology.org/W04-1503

A Simple String-Rewriting Formalism for Dependency Grammar

Recently, dependency grammar has gained renewed attention as empirical methods in parsing have emphasized the importance of relations between words, which is what dependency grammars model explicitly, but context-free phrase-structure grammars do not. While there has been much work on formalizing dependency grammar and on parsing algorithms for dependency grammars in the past, there is not a complete generative formalization of dependency grammar based on string-rewriting in which the derivation structure is the desired dependency structure. Such a system allows for the definition of a compact parse forest in a straightforward manner. In this paper, we present a simple generative formalism for dependency grammars based on Extended Context-Free Grammar, along with a parser; the formalism captures the intuitions of previous formalizations while deviating minimally from the much-used Context-Free Grammar.

2004

zhao-etal-2010-automatic

https://aclanthology.org/C10-2171

Automatic Temporal Expression Normalization with Reference Time Dynamic-Choosing

Temporal expressions in texts contain significant temporal information. Understanding temporal information is very useful in many NLP applications, such as information extraction, documents summarization and question answering. Therefore, the temporal expression normalization which is used for transforming temporal expressions to temporal information has absorbed many researchers' attentions. But previous works, whatever the hand-crafted rules-based or the machine-learnt rules-based, all can not address the actual problem about temporal reference in real texts effectively. More specifically, the reference time choosing mechanism employed by these works is not adaptable to the universal implicit times in normalization. Aiming at this issue, we introduce a new reference time choosing mechanism for temporal expression normalization, called reference time dynamic-choosing, which assigns the appropriate reference times to different classes of implicit temporal expressions dynamically when normalizing. And then, the solution to temporal expression defuzzification by scenario dependences among temporal expressions is discussed. Finally, we evaluate the system on a substantial corpus collected by Chinese news articles and obtained more promising results than compared methods.

2010

dolan-etal-2004-unsupervised

https://aclanthology.org/C04-1051

Unsupervised Construction of Large Paraphrase Corpora: Exploiting Massively Parallel News Sources

We investigate unsupervised techniques for acquiring monolingual sentence-level paraphrases from a corpus of temporally and topically clustered news articles collected from thousands of web-based news sources. Two techniques are employed: (1) simple string edit distance, and (2) a heuristic strategy that pairs initial (presumably summary) sentences from different news stories in the same cluster. We evaluate both datasets using a word alignment algorithm and a metric borrowed from machine translation. Results show that edit distance data is cleaner and more easily-aligned than the heuristic data, with an overall alignment error rate (AER) of 11.58% on a similarly-extracted test set. On test data extracted by the heuristic strategy, however, performance of the two training sets is similar, with AERs of 13.2% and 14.7% respectively. Analysis of 100 pairs of sentences from each set reveals that the edit distance data lacks many of the complex lexical and syntactic alternations that characterize monolingual paraphrase. The summary sentences, while less readily alignable, retain more of the non-trivial alternations that are of greatest interest learning paraphrase relationships.

We are grateful to the Mo Corston-Oliver, Jeff Stevenson and Amy Muia of the Butler Hill Group for their work in annotating the data used in the experiments. We have also benefited from discussions with Ken Church, Mark Johnson, Daniel Marcu and Franz Och. We remain, however, responsible for all content.

2004

ghosh-etal-2020-cease

https://aclanthology.org/2020.lrec-1.201

CEASE, a Corpus of Emotion Annotated Suicide notes in English

A suicide note is usually written shortly before the suicide, and it provides a chance to comprehend the self-destructive state of mind of the deceased. From a psychological point of view, suicide notes have been utilized for recognizing the motive behind the suicide. To the best of our knowledge, there are no openly accessible suicide note corpus at present, making it challenging for the researchers and developers to deep dive into the area of mental health assessment and suicide prevention. In this paper, we create a fine-grained emotion annotated corpus (CEASE) of suicide notes in English, and develop various deep learning models to perform emotion detection on the curated dataset. The corpus consists of 2393 sentences from around 205 suicide notes collected from various sources. Each sentence is annotated with a particular emotion class from a set of 15 fine-grained emotion labels, namely (forgiveness, happiness peacefulness, love, pride, hopefulness, thankfulness, blame, anger, fear, abuse, sorrow, hopelessness, guilt, information, instructions). For the evaluation, we develop an ensemble architecture, where the base models correspond to three supervised deep learning models, namely Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU) and Long Short Term Memory (LSTM).We obtain the highest test accuracy of 60.17%, and cross-validation accuracy of 60.32%.

Good Health and Well-Being

Authors gratefully acknowledge the support from the project titled 'Development of C-DAC Digital Forensic Centre with AI based Knowledge Support Tools', supported by MeitY, Govt. of India and Govt. of Bihar. The authors would also like to thank the linguists: Akash Bhagat, Suman Shekhar (IIT Patna) and Danish Armaan (IIEST Shibpur) for their valuable efforts in labelling the tweets.

2020

horacek-2013-justifying

https://aclanthology.org/R13-1040

Justifying Corpus-Based Choices in Referring Expression Generation

Most empirically-based approaches to NL generation elaborate on co-occurrences and frequencies observed over a corpus, which are then accommodated by learning algorithms. This method fails to capture generalities in generation subtasks, such as generating referring expressions, so that results obtained for some corpus cannot be transferred with confidence to similar environments or even to other domains. In order to obtain a more general basis for choices in referring expression generation, we formulate situational and task-specific properties, and we test to what degree they hold in a specific corpus. As a novelty, we incorporate features of the role of the underlying task, object identification, into these property specifications; these features are inherently domain-independent. Our method has the potential to enable the development of a repertoire of regularities that express generalities and differences across situations and domains, which supports the development of generic algorithms and also leads to a better understanding of underlying dependencies.

2013

carpenter-qu-1995-abstract

https://aclanthology.org/1995.iwpt-1.9

An Abstract Machine for Attribute-Value Logics

A direct abstract machine implementation of the core attribute-value logic operations is shown to decrease the number of operations and conserve the amount of storage required when compared to interpreters or indirect compilers. In this paper, we describe the fundamental data structures and compilation techniques that we have employed to develop a unification and constraint-resolution engine capable of performance rivaling that of directly compiled Prolog terms while greatly exceeding Prolog in flexibility, expressiveness and modularity. In this paper, we will discuss the core architecture of our machine. We begin with a survey of the data structures supporting the small set of attribute-value logic instructions. These instructions manipulate feature structures by means of features, equality, and typing, and manipulate the program state by search and sequencing operations. We further show how these core operations can be integrated with a broad range of standard parsing techniques. Feature structures improve upon Prolog terms by allowing data to be organized by feature rather than by position. This encourages modular program development through the use of sparse structural descriptions which can be logically conjoined into larger units and directly executed. Standard linguistic representations, even of relatively simple local syntactic and semantic structures, typically run to hundreds of substructures. The type discipline we impose organizes information in an object-oriented manner by the multiple inheritance of classes and their associated features and type value constraints. In practice, this allows the construction of large-scale grammars in a relatively short period of time. At run-time, eager copying and structure-sharing is replaced with lazy, incremental, and localized branch and write operations. In order to allow for applications with parallel search, incremental backtracking can be localized to disjunctive choice points within the description of a single structure, thus supporting the kind of conditional mutual consistency checks used in modern grammatical theories such as HPSG, GB, and LFG. Further attention is paid to the byte-coding of instructions and their efficient indexing and subsequent retrieval, all of which is keyed on type information. 1 Motivation Modern attribute-value constraint-based grammars share their primary operational structure with logic programs. In the past decade, Prolog compilers, such as Warren's Abstract Machine (A1t-Kaci 1990), have supplanted interpreters as the execution method of choice for logic pro grams. This is in large part due to a 50-fold speed up in execution times and a reduction by an order of magnitude in terms of space required. In addition to efficiency, compilation also brings the opportunity for static error detection. The vast majority of the time and space used by traditional unification-based grammar interpreters is spent on copying and unifying feature structures. For example, in a bottom-up chart parser, the standard process would be first to build a feature structure for a lexical entry, then to build the feature structures for the relevant rules, and then to unify the matching structures. The principal drawback to this, approach is that complete feature structures have to be constructed, even though unification may result in failure. In the case of f ailure, this can amount to a substantial amount of wasted time and space. By adopting an incremental compiled approach, a description is compiled into a set of abstract machine instructions. At run-time a description is evaluated incrementally, one instruction at a time. In this way, conflicts can be detected as early as possible, before any irrelevant structure has been introduced. In practice, this often means that the inconsistency of a rule with a category can often be detected very

1995

batra-etal-2021-building

https://aclanthology.org/2021.emnlp-main.53

Building Adaptive Acceptability Classifiers for Neural NLG

We propose a novel framework to train models to classify acceptability of responses generated by natural language generation (NLG) models, improving upon existing sentence transformation and model-based approaches. An NLG response is considered acceptable if it is both semantically correct and grammatical. We don't make use of any human references making the classifiers suitable for runtime deployment. Training data for the classifiers is obtained using a 2-stage approach of first generating synthetic data using a combination of existing and new model-based approaches followed by a novel validation framework to filter and sort the synthetic data into acceptable and unacceptable classes. Our 2stage approach adapts to a wide range of data representations and does not require additional data beyond what the NLG models are trained on. It is also independent of the underlying NLG model architecture, and is able to generate more realistic samples close to the distribution of the NLG model-generated responses. We present results on 5 datasets (WebNLG, Cleaned E2E, ViGGO, Alarm, and Weather) with varying data representations. We compare our framework with existing techniques that involve synthetic data generation using simple sentence transformations and/or modelbased techniques, and show that building acceptability classifiers using data that resembles the generation model outputs followed by a validation framework outperforms the existing techniques, achieving state-of-the-art results. We also show that our techniques can be used in few-shot settings using self-training.

2021

husain-etal-2011-clausal

https://aclanthology.org/I11-1143

Clausal parsing helps data-driven dependency parsing: Experiments with Hindi

This paper investigates clausal data-driven dependency parsing. We first motivate a clause as the minimal parsing unit by correlating inter-and intra-clausal relations with relation type, depth, arc length and non-projectivity. This insight leads to a two-stage formulation of parsing where intra-clausal relations are identified in the 1 st stage and inter-clausal relations are identified in the 2 nd stage. We compare two ways of implementing this idea, one based on hard constraints (similar to the one used in constraint-based parsing) and one based on soft constraints (using a kind of parser stacking). Our results show that the approach using hard constraints seems most promising and performs significantly better than single-stage parsing. Our best result gives significant increase in LAS and UAS, respectively, over the previous best result using single-stage parsing.

2011

ws-2001-adaptation

https://aclanthology.org/W01-0300

Adaptation in Dialog Systems

2001

maier-kallmeyer-2010-discontinuity

https://aclanthology.org/W10-4415

Discontinuity and Non-Projectivity: Using Mildly Context-Sensitive Formalisms for Data-Driven Parsing

We present a parser for probabilistic Linear Context-Free Rewriting Systems and use it for constituency and dependency treebank parsing. The choice of LCFRS, a formalism with an extended domain of locality, enables us to model discontinuous constituents and nonprojective dependencies in a straightforward way. The parsing results show that, firstly, our parser is efficient enough to be used for datadriven parsing and, secondly, its result quality for constituency parsing is comparable to the output quality of other state-of-the-art results, all while yielding structures that display discontinuous dependencies.

2010

kiela-bottou-2014-learning

https://aclanthology.org/D14-1005

Learning Image Embeddings using Convolutional Neural Networks for Improved Multi-Modal Semantics

We construct multi-modal concept representations by concatenating a skip-gram linguistic representation vector with a visual concept representation vector computed using the feature extraction layers of a deep convolutional neural network (CNN) trained on a large labeled object recognition dataset. This transfer learning approach brings a clear performance gain over features based on the traditional bag-of-visual-word approach. Experimental results are reported on the WordSim353 and MEN semantic relatedness evaluation tasks. We use visual features computed using either ImageNet or ESP Game images.

We would like to thank Maxime Oquab for providing the feature extraction code.

2014

kyle-etal-2013-native

https://aclanthology.org/W13-1731

Native Language Identification: A Key N-gram Category Approach

This study explores the efficacy of an approach to native language identification that utilizes grammatical, rhetorical, semantic, syntactic, and cohesive function categories comprised of key n-grams. The study found that a model based on these categories of key n-grams was able to successfully predict the L1 of essays written in English by L2 learners from 11 different L1 backgrounds with an accuracy of 59%. Preliminary findings concerning instances of crosslinguistic influence are discussed, along with evidence of language similarities based on patterns of language misclassification.

We thank ETS for compiling and providing the TOEFL11 corpus, and we also thank the organizers of the NLI Shared Task 2013.

2013

bingel-etal-2016-extracting

https://aclanthology.org/P16-1071

Extracting token-level signals of syntactic processing from fMRI - with an application to PoS induction

Neuro-imaging studies on reading different parts of speech (PoS) report somewhat mixed results, yet some of them indicate different activations with different PoS. This paper addresses the difficulty of using fMRI to discriminate between linguistic tokens in reading of running text because of low temporal resolution. We show that once we solve this problem, fMRI data contains a signal of PoS distinctions to the extent that it improves PoS induction with error reductions of more than 4%.

Good Health and Well-Being

This research was partially funded by the ERC Starting Grant LOWLANDS No. 313695, as well as by Trygfonden.

2016

long-etal-2017-xjnlp

https://aclanthology.org/S17-2178

XJNLP at SemEval-2017 Task 12: Clinical temporal information ex-traction with a Hybrid Model

Temporality is crucial in understanding the course of clinical events from a patient's electronic health records and temporal processing is becoming more and more important for improving access to content. SemEval 2017 Task 12 (Clinical TempEval) addressed this challenge using the THYME corpus, a corpus of clinical narratives annotated with a schema based on TimeML2 guidelines. We developed and evaluated approaches for: extraction of temporal expressions (TIMEX3) and EVENTs; EVENT attributes; document-time relations. Our approach is a hybrid model which is based on rule based methods, semi-supervised learning, and semantic features with addition of manually crafted rules.

Good Health and Well-Being

This work has been supported by "The Fundamental Theory and Applications of Big Data with

2017

evans-1996-legitimate

https://aclanthology.org/Y96-1033

Legitimate Termination of Nonlocal Features in HPSG

This paper reviews the treatment of wh-question facts offered by Lappin and Johnson 1996, and suggests that their account of certain island phenomena should be adapted by assuming that certain phrase structures license binding of inherited features. In Japanese, Lappin and Johnson's INHERILQUE feature appears to be dependent on INHERIQUE in order to terminate with a functional C head's TO-BINDIQUE. For certain languages, C's TO-BINDILQUE feature must be null if TO-BINDIQUE is null. In the spirit of Sag 1996 and Pollard and Yoo 1996, the facts can be handled by saying that TO-BINDILQUE is licensed on a wh-clause (wh-cl). As a wh-cl requires TO-BINDIQUE, the dependence of the less robust INHERILQUE on INHERIQUE is thus explained.

1996

suzuki-2012-classifying

https://aclanthology.org/W12-5307

Classifying Hotel Reviews into Criteria for Review Summarization

Recently, we can refer to user reviews in the shopping or hotel reservation sites. However, with the exponential growth of information of the Internet, it is becoming increasingly difficult for a user to read and understand all the materials from a large-scale reviews. In this paper, we propose a method for classifying hotel reviews written in Japanese into criteria, e.g., location and facilities. Our system firstly extracts words which represent criteria from hotel reviews. The extracted words are classified into 12 criteria classes. Then, for each hotel, each sentence of the guest reviews is classified into criterion classes by using two different types of Naive Bayes classifiers. We performed experiments for estimating accuracy of classifying hotel review into 12 criteria. The results showed the effectiveness of our method and indicated that it can be used for review summarization by guest's criteria.

The authors would like to thank the referees for their comments on the earlier version of this paper. This work was partially supported by The Telecommunications Advancement Foundation.

2012

inaba-1996-computational

https://aclanthology.org/Y96-1029

A Computational Expression of Initial Binary Feet and Surface Ternary Feet in Metrical Theory

Under the strict binary foot parsing (Kager 1993), stray elements may occur between bimoraic feet. The stray element may be associated to the preceding foot or following foot at surface level. Stray element adjunction is the mechanism for achieving surface exhaustivity. Each language has its own unique mecbanism of stray element adjunction in order to achieve surface exhaustivity. In Japanese loanwords, the strict binary initial foot parsing creates stray moras. Inaba's (1996) phonetic experiment shows that the word-medial stray moras associate to preceding feet, and provides evidence for the initial unaccented mora as extrametrical. Since the theoretical points I advance are deeply embedded in other languages, I present a set of possible parameters. Based on the set of parameters, I create a computer program which derives the surface foot structures of input loanwords in Japanese, Fijian, and Ponapean.

1996

silva-etal-2010-top

http://www.lrec-conf.org/proceedings/lrec2010/pdf/136_Paper.pdf

Top-Performing Robust Constituency Parsing of Portuguese: Freely Available in as Many Ways as you Can Get it

In this paper we present LX-Parser, a probabilistic, robust constituency parser for Portuguese. This parser achieves ca. 88% f-score in the labeled bracketing task, thus reaching a state-of-the-art performance score that is in line with those that are currently obtained by top-ranking parsers for English, the most studied natural language. To the best of our knowledge, LX-Parser is the first state-of-the-art, robust constituency parser for Portuguese that is made freely available. This parser is being distributed in a variety of ways, each suited for a different type of usage. More specifically, LX-Parser is being made available (i) as a downloadable, stand-alone parsing tool that can be run locally by its users; (ii) as a Web service that exposes an interface that can be invoked remotely and transparently by client applications; and finally (iii) as an on-line parsing service, aimed at human users, that can be accessed through any common Web browser.

2010

perez-beltrachini-lapata-2021-models

https://aclanthology.org/2021.emnlp-main.742

Models and Datasets for Cross-Lingual Summarisation

We present a cross-lingual summarisation corpus with long documents in a source language associated with multi-sentence summaries in a target language. The corpus covers twelve language pairs and directions for four European languages, namely Czech, English, French and German, and the methodology for its creation can be applied to several other languages. We derive cross-lingual document-summary instances from Wikipedia by combining lead paragraphs and articles' bodies from language aligned Wikipedia titles. We analyse the proposed cross-lingual summarisation task with automatic metrics and validate it with a human study. To illustrate the utility of our dataset we report experiments with multilingual pretrained models in supervised, zero-and fewshot, and out-of-domain scenarios.

We thank the anonymous reviewers for their feedback. We also thank Yumo Xu for useful discussions about the models. We are extremely grateful to our bilingual annotators and to Voxeurop SCE publishers. We gratefully acknowledge the support of the European Research Council (award number 681760).

2021

komninos-manandhar-2016-dependency

https://aclanthology.org/N16-1175

Dependency Based Embeddings for Sentence Classification Tasks

We compare different word embeddings from a standard window based skipgram model, a skipgram model trained using dependency context features and a novel skipgram variant that utilizes additional information from dependency graphs. We explore the effectiveness of the different types of word embeddings for word similarity and sentence classification tasks. We consider three common sentence classification tasks: question type classification on the TREC dataset, binary sentiment classification on Stanford's Sentiment Treebank and semantic relation classification on the SemEval 2010 dataset. For each task we use three different classification methods: a Support Vector Machine, a Convolutional Neural Network and a Long Short Term Memory Network. Our experiments show that dependency based embeddings outperform standard window based embeddings in most of the settings, while using dependency context embeddings as additional features improves performance in all tasks regardless of the classification method. Our embeddings and code are available at https://www.cs.york.ac.uk/nlp/ extvec

Alexandros Komninos was supported by EP-SRC via an Engineering Doctorate in LSCITS. Suresh Manandhar was supported by EPSRC grant EP/I037512/1, A Unified Model of Compositional & Distributional Semantics: Theory and Application.

2016

merlo-van-der-plas-2009-abstraction

https://aclanthology.org/P09-1033

Abstraction and Generalisation in Semantic Role Labels: PropBank, VerbNet or both?

Semantic role labels are the representation of the grammatically relevant aspects of a sentence meaning. Capturing the nature and the number of semantic roles in a sentence is therefore fundamental to correctly describing the interface between grammar and meaning. In this paper, we compare two annotation schemes, Prop-Bank and VerbNet, in a task-independent, general way, analysing how well they fare in capturing the linguistic generalisations that are known to hold for semantic role labels, and consequently how well they grammaticalise aspects of meaning. We show that VerbNet is more verb-specific and better able to generalise to new semantic role instances, while PropBank better captures some of the structural constraints among roles. We conclude that these two resources should be used together, as they are complementary.

We thank James Henderson and Ivan Titov for useful comments. The research leading to these results has received partial funding from the EU FP7 programme (FP7/2007-2013) under grant agreement number 216594 (CLASSIC project: www.classic-project.org).

2009

green-2011-effects

https://aclanthology.org/P11-3013

Effects of Noun Phrase Bracketing in Dependency Parsing and Machine Translation

Flat noun phrase structure was, up until recently, the standard in annotation for the Penn Treebanks. With the recent addition of internal noun phrase annotation, dependency parsing and applications down the NLP pipeline are likely affected. Some machine translation systems, such as TectoMT, use deep syntax as a language transfer layer. It is proposed that changes to the noun phrase dependency parse will have a cascading effect down the NLP pipeline and in the end, improve machine translation output, even with a reduction in parser accuracy that the noun phrase structure might cause. This paper examines this noun phrase structure's effect on dependency parsing, in English, with a maximum spanning tree parser and shows a 2.43%, 0.23 Bleu score, improvement for English to Czech machine translation.

This research has received funding from the European Commissions 7th Framework Program (FP7) under grant agreement n • 238405 (CLARA), and from grant MSM 0021620838. I would like to thank ZdeněkŽabokrtský for his guidance in this research and also the anonymous reviewers for their comments.

2011

fei-etal-2020-mimic

https://aclanthology.org/2020.findings-emnlp.18

Mimic and Conquer: Heterogeneous Tree Structure Distillation for Syntactic NLP

Syntax has been shown useful for various NLP tasks, while existing work mostly encodes singleton syntactic tree using one hierarchical neural network. In this paper, we investigate a simple and effective method, Knowledge Distillation, to integrate heterogeneous structure knowledge into a unified sequential LSTM encoder. Experimental results on four typical syntax-dependent tasks show that our method outperforms tree encoders by effectively integrating rich heterogeneous structure syntax, meanwhile reducing error propagation, and also outperforms ensemble methods, in terms of both the efficiency and accuracy.

This work is supported by the National Natural Science Foundation of China (No. 61772378, No. 61702121)

2020

mcintosh-2009-canadian

https://aclanthology.org/2009.mtsummit-government.4

Canadian Job Bank Automated Translation System

§ Job Bank (www.jobbank.gc.ca) is a free job-posting service provided by the Federal Government to all Canadians. Employers have the option to create a profile; upon approval, they can then post job offers. § Job seekers are able to access these positions in two ways. Standard Job Search and Job Matching § Several additional tools are available to assist the Job Seeker with their job search; such as resume builder, job alert, job search tips, and career navigator Job Bank for Employers § Employers can post job advertisements 24 hours a day, 7 days a week using the "Job Bank for Employers" Web site § Job offers received by fax, e-mail, Internet and telephone must be published simultaneously in both French and English within 24 business hours § 70,356,222 Job Bank Web site visits in 2008-2009 § 1,138,233 Spelling and Grammar Checker § Customized entries are added on a weekly basis to a single file § Has been integrated into the JBFE interface Oracle Database § Archives offers and their post-edited equivalents in a database § Automatically posts offers that are identical (100% match) along with their translation § Of all offers posted to the JB site, 45% are reproduced by the database

Decent Work and Economic Growth

2009

sornlertlamvanich-etal-2000-automatic

https://aclanthology.org/C00-2116

Automatic Corpus-Based Thai Word Extraction with the C4.5 Learning Algorithm

Word" is difficult to define in the languages that do not exhibit explicit word boundary, such as Thai. Traditional methods on defining words for this kind of languages have to depend on human judgement which bases on unclear criteria or procedures, and have several limitations. This paper proposes an algorithm for word extraction from Thai texts without borrowing a hand from word segmentation. We employ the c4.5 learning algorithm for this task. Several attributes such as string length, frequency, mutual information and entropy are chosen for word/non-word determination. Our experiment yields high precision results about 85% in both training and test corpus.

Special thanks to Assistant Professor Mikio Yamamoto for providing the useful program to extract all substrings from the corpora in linear time.

2000

song-etal-2020-multi

https://aclanthology.org/2020.emnlp-main.546

Multi-Stage Pre-training for Automated Chinese Essay Scoring

This paper proposes a pre-training based automated Chinese essay scoring method. The method involves three components: weakly supervised pre-training, supervised crossprompt fine-tuning and supervised targetprompt fine-tuning. An essay scorer is first pretrained on a large essay dataset covering diverse topics and with coarse ratings, i.e., good and poor, which are used as a kind of weak supervision. The pre-trained essay scorer would be further fine-tuned on previously rated essays from existing prompts, which have the same score range with the target prompt and provide extra supervision. At last, the scorer is fine-tuned on the target-prompt training data. The evaluation on four prompts shows that this method can improve a state-of-the-art neural essay scorer in terms of effectiveness and domain adaptation ability, while in-depth analysis also reveals its limitations.

Quality Education

This work is supported by the National Natural Science Foundation of China (Nos. 61876113, 61876112), Beijing Natural Science Foundation (No. 4192017) and Capital Building for Sci-Tech Innovation-Fundamental Scientific Research Funds. Lizhen Liu is the corresponding author.

2020

vu-etal-2022-domain

https://aclanthology.org/2022.findings-acl.49

Domain Generalisation of NMT: Fusing Adapters with Leave-One-Domain-Out Training

Generalising to unseen domains is underexplored and remains a challenge in neural machine translation. Inspired by recent research in parameter-efficient transfer learning from pretrained models, this paper proposes a fusionbased generalisation method that learns to combine domain-specific parameters. We propose a leave-one-domain-out training strategy to avoid information leaking to address the challenge of not knowing the test domain during training time. Empirical results on three language pairs show that our proposed fusion method outperforms other baselines up to +0.8 BLEU score on average.

This research is supported by an eBay Research Award and the ARC Future Fellowship FT190100039. This work is partly sponsored by the Air Force Research Laboratory and DARPA under agreement number FA8750-19-2-0501. The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation thereon. The authors are grateful to the anonymous reviewers for their helpful comments to improve the manuscript.

2022

bisk-hockenmaier-2013-hdp

https://aclanthology.org/Q13-1007

An HDP Model for Inducing Combinatory Categorial Grammars

We introduce a novel nonparametric Bayesian model for the induction of Combinatory Categorial Grammars from POS-tagged text. It achieves state of the art performance on a number of languages, and induces linguistically plausible lexicons.

This work is supported by NSF CAREER award 1053856 (Bayesian Models for Lexicalized Grammars).

2013

jauhiainen-etal-2017-evaluating

https://aclanthology.org/W17-1212

Evaluating HeLI with Non-Linear Mappings

In this paper we describe the non-linear mappings we used with the Helsinki language identification method, HeLI, in the 4 th edition of the Discriminating between Similar Languages (DSL) shared task, which was organized as part of the Var-Dial 2017 workshop. Our SUKI team participated in the closed track together with 10 other teams. Our system reached the 7 th position in the track. We describe the HeLI method and the non-linear mappings in mathematical notation. The HeLI method uses a probabilistic model with character n-grams and word-based backoff. We also describe our trials using the non-linear mappings instead of relative frequencies and we present statistics about the back-off function of the HeLI method.

We would like to thank Kimmo Koskenniemi for many valuable discussions and comments. This research was made possible by funding from the Kone Foundation Language Programme.

2017

zhang-etal-2020-fast

https://aclanthology.org/2020.wmt-1.62

Fast Interleaved Bidirectional Sequence Generation

Independence assumptions during sequence generation can speed up inference, but parallel generation of highly interdependent tokens comes at a cost in quality. Instead of assuming independence between neighbouring tokens (semi-autoregressive decoding, SA), we take inspiration from bidirectional sequence generation and introduce a decoder that generates target words from the left-to-right and right-toleft directions simultaneously. We show that we can easily convert a standard architecture for unidirectional decoding into a bidirectional decoder by simply interleaving the two directions and adapting the word positions and selfattention masks. Our interleaved bidirectional decoder (IBDecoder) retains the model simplicity and training efficiency of the standard Transformer, and on five machine translation tasks and two document summarization tasks, achieves a decoding speedup of ∼2× compared to autoregressive decoding with comparable quality. Notably, it outperforms left-toright SA because the independence assumptions in IBDecoder are more felicitous. To achieve even higher speedups, we explore hybrid models where we either simultaneously predict multiple neighbouring tokens per direction, or perform multi-directional decoding by partitioning the target sequence. These methods achieve speedups to 4×-11× across different tasks at the cost of <1 BLEU or <0.5 ROUGE (on average). 1

This work was performed using resources provided by the Cambridge Service for Data Driven Discovery (CSD3) operated by the University of Cambridge Research Computing Service (http: //www.csd3.cam.ac.uk/), provided by Dell EMC and Intel using Tier-2 funding from the Engineering and Physical Sciences Research Council (capital grant EP/P020259/1), and DiRAC funding from the Science and Technology Facilities Council (www.dirac.ac.uk). Ivan Titov acknowledges support of the European Research Council (ERC Starting grant 678254) and the Dutch National Science Foundation (NWO VIDI 639.022.518). Rico Sennrich acknowledges support of the Swiss National Science Foundation (MUTAMUR; no. 176727).

2020

grefenstette-2015-inriasac

https://aclanthology.org/S15-2152

INRIASAC: Simple Hypernym Extraction Methods

For information retrieval, it is useful to classify documents using a hierarchy of terms from a domain. One problem is that, for many domains, hierarchies of terms are not available. The task 17 of SemEval 2015 addresses the problem of structuring a set of terms from a given domain into a taxonomy without manual intervention. Here we present some simple taxonomy structuring techniques, such as term overlap and document and sentence cooccurrence in large quantities of text (English Wikipedia) to produce hypernym pairs for the eight domain lists supplied by the task organizers. Our submission ranked first in this 2015 benchmark, which suggests that overly complicated methods might need to be adapted to individual domains. We describe our generic techniques and present an initial evaluation of results.

This research is partially funded by a research grant from INRIA, and the Paris-Saclay Institut de la Société Numérique funded by the IDEX Paris-Saclay, ANR-11-IDEX-0003-02.

2015

seyffarth-kallmeyer-2020-corpus

https://aclanthology.org/2020.coling-main.357

Corpus-based Identification of Verbs Participating in Verb Alternations Using Classification and Manual Annotation

English verb alternations allow participating verbs to appear in a set of syntactically different constructions whose associated semantic frames are systematically related. We use ENCOW and VerbNet data to train classifiers to predict the instrument subject alternation and the causativeinchoative alternation, relying on count-based and vector-based features as well as perplexitybased language model features, which are intended to reflect each alternation's felicity by simulating it. Beyond the prediction task, we use the classifier results as a source for a manual annotation step in order to identify new, unseen instances of each alternation. This is possible because existing alternation datasets contain positive, but no negative instances and are not comprehensive. Over several sequences of classification-annotation steps, we iteratively extend our sets of alternating verbs. Our hybrid approach to the identification of new alternating verbs reduces the required annotation effort by only presenting annotators with the highest-scoring candidates from the previous classification. Due to the success of semi-supervised and unsupervised features, our approach can easily be transferred to further alternations.

The work presented in this paper was financed by the Deutsche Forschungsgemeinschaft (DFG) within the CRC 991 "The Structure of Representations in Language, Cognition, and Science" and the individual DFG project "Unsupervised Frame Induction (FInd)". We wish to thank the anonymous reviewers for their constructive feedback and helpful comments.

2020

shen-etal-2021-sciconceptminer

https://aclanthology.org/2021.acl-demo.6

SciConceptMiner: A system for large-scale scientific concept discovery

Scientific knowledge is evolving at an unprecedented rate of speed, with new concepts constantly being introduced from millions of academic articles published every month. In this paper, we introduce a self-supervised end-toend system, SciConceptMiner, for the automatic capture of emerging scientific concepts from both independent knowledge sources (semi-structured data) and academic publications (unstructured documents). First, we adopt a BERT-based sequence labeling model to predict candidate concept phrases with selfsupervision data. Then, we incorporate rich Web content for synonym detection and concept selection via a web search API. This two-stage approach achieves highly accurate (94.7%) concept identification with more than 740K scientific concepts. These concepts are deployed in the Microsoft Academic 1 production system and are the backbone for its semantic search capability.

Industry, Innovation and Infrastructure

13 We split the sampled data of each category to 3 groups with 100 each and they are evaluated by 3 judges. We report the average of positive label ratios.Source Age Avg Y 5% Y 50% Y 95% Y Wiki

2021

niu-2017-chinese

https://aclanthology.org/W17-6519

Chinese Descriptive and Resultative V-de Constructions. A Dependency-based Analysis

This contribution presents a dependency grammar (DG) analysis of the so-called descriptive and resultative V-de constructions in Mandarin Chinese (VDCs); it focuses, in particular, on the dependency analysis of the noun phrase that intervenes between the two predicates in a VDC. Two methods, namely chunking data collected from informants and two diagnostics specific to Chinese, i.e. bǎ and bèi sentence formation, were used. They were employed to discern which analysis should be preferred, i.e. the ternary-branching analysis, in which the intervening NP (NP2) is a dependent of the first predicate (P1), or the small-clause analysis, in which NP2 depends on the second predicate (P2). The results obtained suggest a flexible structural analysis for VDCs in the form of "NP1+P1-de+NP2+P2". The difference in structural assignment is attributed to a semantic property of NP2 and the semantic relations it forms with adjacent predicates.

The research presented in this article was funded by the Ministry of Education of the People's Republic of China, Grant # 15YJA74001.

2017

lichouri-abbas-2020-speechtrans

https://aclanthology.org/2020.smm4h-1.19

SpeechTrans@SMM4H'20: Impact of Preprocessing and N-grams on Automatic Classification of Tweets That Mention Medications

This paper describes our system developed for automatically classifying tweets that mention medications. We used the Decision Tree classifier for this task. We have shown that using some elementary preprocessing steps and TF-IDF n-grams led to acceptable classifier performance. Indeed, the F1-score recorded was 74.58% in the development phase and 63.70% in the test phase.

Good Health and Well-Being

2020

alexin-etal-2003-annotated

https://aclanthology.org/E03-1012

Annotated Hungarian National Corpus

The beginning of the work dates back to 1998 when the authors started a research project on the application of ILP (Inductive Logic Programming) learning methods for part-of-speech tagging. This research was done within the framework of a European ESPRIT project (LTR 20237, "lLP2") , where first studies were based on the so-called TELRI corpus (Erjavec et al., 1998) . Since the corpus annotation had several deficiencies and its size proved to be small for further research, a national project has been organized with the main goal to create a suitably large training corpus for machine learning applications, primarily for POS (Part-of-speech) tagging.

The project was partially supported by the Hungarian Ministry of Education (grant: IKTA 27/2000). The authors also would like to thank researchers of the Research Institute for Linguistics at the Hungarian Academy of Sciences for their kind help and advice.

2003

meyers-etal-2004-cross

http://www.lrec-conf.org/proceedings/lrec2004/pdf/397.pdf

The Cross-Breeding of Dictionaries

Especially for English, the number of hand-coded electronic resources available to the Natural Language Processing Community keeps growing: annotated corpora, treebanks, lexicons, wordnets, etc. Unfortunately, initial funding for such projects is much easier to obtain than the additional funding needed to enlarge or improve upon such resources. Thus once one proves the usefulness of a resource, it is difficult to make that resource reach its full potential. We discuss techniques for combining dictionary resources and producing others by semi-automatic means. The resources we created using these techniques have become an integral part of our work on NomBank, a project with the goal of annotating noun arguments in the Penn Treebank II corpus (PTB).

Nombank is supported under Grant N66001-001-1-8917 from the Space and Naval Warfare Systems Center San Diego. This paper does not necessarily reflect the position or the policy of the U.S. Government.

2004

janarthanam-lemon-2010-adaptive

https://aclanthology.org/W10-4324

Adaptive Referring Expression Generation in Spoken Dialogue Systems: Evaluation with Real Users

We present new results from a real-user evaluation of a data-driven approach to learning user-adaptive referring expression generation (REG) policies for spoken dialogue systems. Referring expressions can be difficult to understand in technical domains where users may not know the technical 'jargon' names of the domain entities. In such cases, dialogue systems must be able to model the user's (lexical) domain knowledge and use appropriate referring expressions. We present a reinforcement learning (RL) framework in which the system learns REG policies which can adapt to unknown users online. For real users of such a system, we show that in comparison to an adaptive hand-coded baseline policy, the learned policy performs significantly better, with a 20.8% average increase in adaptation accuracy, 12.6% decrease in time taken, and a 15.1% increase in task completion rate. The learned policy also has a significantly better subjective rating from users. This is because the learned policies adapt online to changing evidence about the user's domain expertise. We also discuss the issue of evaluation in simulation versus evaluation with real users.

The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement no. 216594 (CLASSiC project www.classic-project.org) and from the EPSRC, project no. EP/G069840/1.

2010

sitbon-bellot-2006-tools

http://www.lrec-conf.org/proceedings/lrec2006/pdf/410_pdf.pdf

Tools and methods for objective or contextual evaluation of topic segmentation

In this paper we discuss the way of evaluating topic segmentation, from mathematical measures on variously constructed reference corpus to contextual evaluation depending on different topic segmentation usages. We present an overview of the different ways of building reference corpora and of mathematically evaluating segmentation methods, and then we focus on three tasks which may involve a topic segmentation : text extraction, information retrieval and document presentation. We have developped two graphical interfaces, one for an intrinsec comparison, and the other one dedicated to an evaluation in an information retrieval context. These tools will be very soon distributed under GPL licences on the Technolangue project web page.

2006

foret-nir-2002-rigid

https://aclanthology.org/C02-1111

Rigid Lambek Grammars Are Not Learnable from Strings

2002

defauw-etal-2019-collecting

https://aclanthology.org/W19-6733

Collecting domain specific data for MT: an evaluation of the ParaCrawlpipeline

This paper investigates the effectiveness of the ParaCrawl pipeline for collecting domain-specific training data for machine translation. We follow the different steps of the pipeline (document alignment, sentence alignment, cleaning) and add a topic-filtering component. Experiments are performed on the legal domain for the English to French and English to Irish language pairs. We evaluate the pipeline at both intrinsic (alignment quality) and extrinsic (MT performance) levels. Our results show that with this pipeline we obtain highquality alignments and significant improvements in MT quality.

This work was performed in the framework of the SMART 2015/1091 project ("Tools and resources for CEF automated translation"), funded by the CEF Telecom programme (Connecting Europe Facility).

2019

jonnalagadda-etal-2013-evaluating

https://aclanthology.org/W13-0404

Evaluating the Use of Empirically Constructed Lexical Resources for Named Entity Recognition

One of the most time-consuming tasks faced by a Natural Language Processing (NLP) researcher or practitioner trying to adapt a machine-learning-based NER system to a different domain is the creation, compilation, and customization of the needed lexicons. Lexical resources, such as lexicons of concept classes are considered necessary to improve the performance of NER. It is typical for medical informatics researchers to implement modularized systems that cannot be generalized (Stanfill et al. 2010) . As the work of constructing or customizing lexical resources needed for these highly specific systems is human-intensive, automatic generation is a desirable alternative. It might be possible that empirically created lexical resources might incorporate domain knowledge into a machine-learning NER engine and increase its accuracy. Although many machine learning-based NER techniques require annotated data, semi-supervised and unsupervised techniques for NER have been long been explored due to their value in domain robustness and minimizing labor costs. Some attempts at automatic knowledgebase construction included automatic thesaurus discovery efforts (Grefenstette 1994) , which sought to build lists of similar words without human intervention to aid in query expansion or automatic dictionary construction (Riloff 1996) . More recently, the use of empirically derived semantics for NER is used by Finkel and Manning (Finkel and Manning 2009a) , Turian et al. (Turian et al. 2010) , and ). Finkel's NER tool uses clusters of terms built apriori from the British National corpus (Aston and Burnard 1998) and English gigaword corpus (Graff et al. 2003) for extracting concepts from newswire text and PubMed abstracts for extracting gene mentions from biomedical literature. Turian et al. (Turian et al. 2010 ) also showed that statistically created word clusters (P. F. Brown et al. 1992; Clark 2000) could be used to improve named entity recognition. However, only a single feature (cluster membership) can be derived from the clusters. Semantic vector representations of terms had not been previously used for NER or sequential tagging classification tasks before (Turian et al. 2010) . Although use empirically derived vector representation for extracting concepts defined in the GENIA (Kim, Ohta, and Tsujii 2008) ontology from biomedical literature using rule-based methods, it was not clear whether such methods could be ported to extract other concepts or incrementally improve the performance of an existing system . This work not only demonstrates how such vector representation could improve state-of-the-art NER, but also that they are more useful than statistical clustering in this context.

This work was possible because of funding from possible sources: NLM HHSN276201000031C (PI: Gonzalez), NCRR 3UL1RR024148, NCRR 1RC1RR028254, NSF 0964613 and the Brown Foundation (PI: Bernstam), NSF ABI:0845523, NLM R01LM009959A1 (PI: Liu) and NLM 1K99LM011389 (PI: Jonnalagadda). We also thank the developers of BANNER (http://banner.sourceforge.net/), MALLET (http://mallet.cs.umass.edu/) and Semantic Vectors (http://code.google.com/p/semanticvectors/) for the software packages and the organizers of the i2b2/VA 2010 NLP challenge for sharing the corpus.

2013

dai-etal-2020-multi

https://aclanthology.org/2020.emnlp-main.565

A Multi-Task Incremental Learning Framework with Category Name Embedding for Aspect-Category Sentiment Analysis

(T)ACSA tasks, including aspect-category sentiment analysis (ACSA) and targeted aspectcategory sentiment analysis (TACSA), aims at identifying sentiment polarity on predefined categories. Incremental learning on new categories is necessary for (T)ACSA real applications. Though current multi-task learning models achieve good performance in (T)ACSA tasks, they suffer from catastrophic forgetting problems in (T)ACSA incremental learning tasks. In this paper, to make multi-task learning feasible for incremental learning, we proposed Category Name Embedding network (CNE-net). We set both encoder and decoder shared among all categories to weaken the catastrophic forgetting problem. Besides the origin input sentence, we applied another input feature, i.e., category name, for task discrimination. Our model achieved state-of-theart on two (T)ACSA benchmark datasets. Furthermore, we proposed a dataset for (T)ACSA incremental learning and achieved the best performance compared with other strong baselines.

2020

nn-1977-finite-string-volume-14-number-5

https://aclanthology.org/J77-3003

The FINITE STRING, Volume 14, Number 5

AMERICAN JOURNAL OF COMPUTATIONAL LINGUISTICS is published by the Association for Computational Linguistics.

1977

tyers-etal-2012-flexible

https://aclanthology.org/2012.eamt-1.54

Flexible finite-state lexical selection for rule-based machine translation

In this paper we describe a module (rule formalism, rule compiler and rule processor) designed to provide flexible support for lexical selection in rule-based machine translation. The motivation and implementation for the system is outlined and an efficient algorithm to compute the best coverage of lexical-selection rules over an ambiguous input sentence is described. We provide a demonstration of the module by learning rules for it on a typical training corpus and evaluating against other possible lexicalselection strategies. The inclusion of the module, along with rules learnt from the parallel corpus provides a small, but consistent and statistically-significant improvement over either using the highest-scoring translation according to a target-language model or using the most frequent aligned translation in the parallel corpus which is also found in the system's bilingual dictionaries.

We are thankful for the support of the Spanish Ministry of Science and Innovation through project TIN2009-14009-C02-01, and the Universitat d'Alacant through project GRE11-20. We also thank Sergio Ortiz Rojas for his constructive comments and ideas on the development of the system, and the anonymous reviewers for comments on the manuscript.

2012

pucher-2007-wordnet

https://aclanthology.org/P07-2033

WordNet-based Semantic Relatedness Measures in Automatic Speech Recognition for Meetings

This paper presents the application of WordNet-based semantic relatedness measures to Automatic Speech Recognition (ASR) in multi-party meetings. Different word-utterance context relatedness measures and utterance-coherence measures are defined and applied to the rescoring of Nbest lists. No significant improvements in terms of Word-Error-Rate (WER) are achieved compared to a large word-based ngram baseline model. We discuss our results and the relation to other work that achieved an improvement with such models for simpler tasks.

This work was supported by the European Union 6th FP IST Integrated Project AMI (Augmented Multiparty Interaction, and by Kapsch Carrier-Com AG and Mobilkom Austria AG together with the Austrian competence centre programme Kplus.

2007

calzolari-etal-2004-enabler

http://www.lrec-conf.org/proceedings/lrec2004/pdf/545.pdf

ENABLER Thematic Network of National Projects: Technical, Strategic and Political Issues of LRs

In this paper we present general strategies concerning Language Resources (LRs)-Written, Spoken and, recently, Multimodal-as developed within the ENABLER Thematic Network. LRs are a central component of the so-called "linguistic infrastructure" (the other key element being Evaluation), necessary for the development of any Human Language Technology (HLT) application. They play a critical role, as horizontal technology, in different emerging areas of FP6, and have been recognized as a priority within a number of national projects around Europe and worldwide. The availability of LRs is also a "sensitive" issue, touching directly the sphere of linguistic and cultural identity, but also with economical, societal and political implications. This is going to be even more true in the new Europe with 25 languages on a par.

Industry, Innovation and Infrastructure

Partnership for the goals

2004

escoter-etal-2017-grouping

https://aclanthology.org/E17-1103

Grouping business news stories based on salience of named entities

In news aggregation systems focused on broad news domains, certain stories may appear in multiple articles. Depending on the relative importance of the story, the number of versions can reach dozens or hundreds within a day. The text in these versions may be nearly identical or quite different. Linking multiple versions of a story into a single group brings several important benefits to the end-user-reducing the cognitive load on the reader, as well as signaling the relative importance of the story. We present a grouping algorithm, and explore several vector-based representations of input documents: from a baseline using keywords, to a method using salience-a measure of importance of named entities in the text. We demonstrate that features beyond keywords yield substantial improvements, verified on a manually-annotated corpus of business news stories.

2017

sartorio-etal-2013-transition

https://aclanthology.org/P13-1014

A Transition-Based Dependency Parser Using a Dynamic Parsing Strategy

We present a novel transition-based, greedy dependency parser which implements a flexible mix of bottom-up and top-down strategies. The new strategy allows the parser to postpone difficult decisions until the relevant information becomes available. The novel parser has a ∼12% error reduction in unlabeled attachment score over an arc-eager parser, with a slowdown factor of 2.8.

We wish to thank Liang Huang and Marco Kuhlmann for discussion related to the ideas reported in this paper, and the anonymous reviewers for their useful suggestions. The second author has been partially supported by MIUR under project PRIN No. 2010LYA9RH 006.

2013

bagga-2000-analyzing

https://aclanthology.org/W00-0106

Analyzing the Reading Comprehension Task

In this paper we describe a method for analyzing the reading comprehension task. First, we describe a method of classifying facts (information) into categories or levels; where each level signifies a different degree of difficulty of extracting a fact from a piece of text containing it. We then proceed to show how one can use this model the analyze the complexity of the reading comprehension task. Finally, we analyze five different reading comprehension tasks and present results from this analysis.

2000

chen-etal-2020-distilling

https://aclanthology.org/2020.acl-main.705

Distilling Knowledge Learned in BERT for Text Generation

Large-scale pre-trained language model such as BERT has achieved great success in language understanding tasks. However, it remains an open question how to utilize BERT for language generation. In this paper, we present a novel approach, Conditional Masked Language Modeling (C-MLM), to enable the finetuning of BERT on target generation tasks. The finetuned BERT (teacher) is exploited as extra supervision to improve conventional Seq2Seq models (student) for better text generation performance. By leveraging BERT's idiosyncratic bidirectional nature, distilling knowledge learned in BERT can encourage auto-regressive Seq2Seq models to plan ahead, imposing global sequence-level supervision for coherent text generation. Experiments show that the proposed approach significantly outperforms strong Transformer baselines on multiple language generation tasks such as machine translation and text summarization. Our proposed model also achieves new state of the art on IWSLT German-English and English-Vietnamese MT datasets. 1

2020

castillo-2010-machine

https://aclanthology.org/W10-1609

A Machine Learning Approach for Recognizing Textual Entailment in Spanish

This paper presents a system that uses machine learning algorithms for the task of recognizing textual entailment in Spanish language. The datasets used include SPARTE Corpus and a translated version to Spanish of RTE3, RTE4 and RTE5 datasets. The features chosen quantify lexical, syntactic and semantic level matching between text and hypothesis sentences. We analyze how the different sizes of datasets and classifiers could impact on the final overall performance of the RTE classification of two-way task in Spanish. The RTE system yields 60.83% of accuracy and a competitive result of 66.50% of accuracy is reported by train and test set taken from SPARTE Corpus with 70% split.

2010

wich-etal-2020-impact

https://aclanthology.org/2020.alw-1.7

Impact of Politically Biased Data on Hate Speech Classification

One challenge that social media platforms are facing nowadays is hate speech. Hence, automatic hate speech detection has been increasingly researched in recent years-in particular with the rise of deep learning. A problem of these models is their vulnerability to undesirable bias in training data. We investigate the impact of political bias on hate speech classification by constructing three politicallybiased data sets (left-wing, right-wing, politically neutral) and compare the performance of classifiers trained on them. We show that (1) political bias negatively impairs the performance of hate speech classifiers and (2) an explainable machine learning model can help to visualize such bias within the training data. The results show that political bias in training data has an impact on hate speech classification and can become a serious issue.

Peace, Justice and Strong Institutions

This paper is based on a joined work in the context of Jan Bauer's master's thesis (Bauer, 2020) . This research has been partially funded by a scholarship from the Hanns Seidel Foundation financed by the German Federal Ministry of Education and Research.

2020

temperley-2010-invited

https://aclanthology.org/N10-1114

Invited Talk: Music, Language, and Computational Modeling: Lessons from the Key-Finding Problem

Recent research in computational music research, including my own, has been greatly influenced by methods in computational linguistics. But I believe the influence could also go the other way: Music may offer some interesting lessons for language research, particularly with regard to the modeling of cognition.

2010

singh-etal-2011-large

https://aclanthology.org/P11-1080

Large-Scale Cross-Document Coreference Using Distributed Inference and Hierarchical Models

Cross-document coreference, the task of grouping all the mentions of each entity in a document collection, arises in information extraction and automated knowledge base construction. For large collections, it is clearly impractical to consider all possible groupings of mentions into distinct entities. To solve the problem we propose two ideas: (a) a distributed inference technique that uses parallelism to enable large scale processing, and (b) a hierarchical model of coreference that represents uncertainty over multiple granularities of entities to facilitate more effective approximate inference. To evaluate these ideas, we constructed a labeled corpus of 1.5 million disambiguated mentions in Web pages by selecting link anchors referring to Wikipedia entities. We show that the combination of the hierarchical model with distributed inference quickly obtains high accuracy (with error reduction of 38%) on this large dataset, demonstrating the scalability of our approach.

This work was done when the first author was an intern at Google Research. The authors would like to thank Mark Dredze, Sebastian Riedel, and anonymous reviewers for their valuable feedback. This work was supported in part by the Center for Intelligent Information Retrieval, the University of Massachusetts gratefully acknowledges the support of Defense Advanced Research Projects Agency (DARPA) Machine Reading Program under Air Force Research Laboratory (AFRL) prime contract no. FA8750-09-C-0181., in part by an award from Google, in part by The Central Intelligence Agency, the National Security Agency and National Science Foundation under NSF grant #IIS-0326249, in part by NSF grant #CNS-0958392, and in part by UPenn NSF medium IIS-0803847. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect those of the sponsor.

2011

dethlefs-etal-2014-cluster

https://aclanthology.org/E14-1074

Cluster-based Prediction of User Ratings for Stylistic Surface Realisation

Surface realisations typically depend on their target style and audience. A challenge in estimating a stylistic realiser from data is that humans vary significantly in their subjective perceptions of linguistic forms and styles, leading to almost no correlation between ratings of the same utterance. We address this problem in two steps. First, we estimate a mapping function between the linguistic features of a corpus of utterances and their human style ratings. Users are partitioned into clusters based on the similarity of their ratings, so that ratings for new utterances can be estimated, even for new, unknown users. In a second step, the estimated model is used to re-rank the outputs of a number of surface realisers to produce stylistically adaptive output. Results confirm that the generated styles are recognisable to human judges and that predictive models based on clusters of users lead to better rating predictions than models based on an average population of users.

Acknowledgements This research was funded by the EC FP7 programme FP7/2011-14 under grant agreements no. 270019 (SPACEBOOK) and no. 287615 (PARLANCE).

2014

yang-etal-2016-extraction

https://aclanthology.org/N16-2012

Extraction of Bilingual Technical Terms for Chinese-Japanese Patent Translation

The translation of patents or scientific papers is a key issue that should be helped by the use of statistical machine translation (SMT). In this paper, we propose a method to improve Chinese-Japanese patent SMT by premarking the training corpus with aligned bilingual multi-word terms. We automatically extract multi-word terms from monolingual corpora by combining statistical and linguistic filtering methods. We use the sampling-based alignment method to identify aligned terms and set some threshold on translation probabilities to select the most promising bilingual multi-word terms. We pre-mark a Chinese-Japanese training corpus with such selected aligned bilingual multi-word terms. We obtain the performance of over 70% precision in bilingual term extraction and a significant improvement of BLEU scores in our experiments on a Chinese-Japanese patent parallel corpus.

Industry, Innovation and Infrastructure

2016

magri-2014-error

https://aclanthology.org/W14-2802

The Error-driven Ranking Model of the Acquisition of Phonotactics: How to Keep the Faithfulness Constraints at Bay

A problem which arises in the theory of the error-driven ranking model of the acquisition of phonotactics is that the faithfulness constraints need to be promoted but should not be promoted too high. This paper motivates this technical problem and shows how to tune the promotion component of the re-ranking rule so as to keep the faithfulness constraints at bay.

This research was supported by a Marie Curie Intra European Fellowship within the 7th European

2014