before_sent
stringlengths 13
1.44k
| before_sent_with_intent
stringlengths 25
1.45k
| after_sent
stringlengths 0
1.41k
| labels
stringclasses 6
values | doc_id
stringlengths 4
10
| revision_depth
int64 1
4
|
|---|---|---|---|---|---|
Recent studies indicate that NLU models are prone to rely on shortcut features for prediction . | <meaning-changed> Recent studies indicate that NLU models are prone to rely on shortcut features for prediction . | Recent studies indicate that NLU models are prone to rely on shortcut features for prediction , without achieving true language understanding . | meaning-changed | 2103.06922 | 1 |
As a result, these models could potentially fail to generalize to real-world out-of-distribution scenarios . | <clarity> As a result, these models could potentially fail to generalize to real-world out-of-distribution scenarios . | As a result, these models fail to generalize to real-world out-of-distribution scenarios . | clarity | 2103.06922 | 1 |
As a result, these models could potentially fail to generalize to real-world out-of-distribution scenarios . | <clarity> As a result, these models could potentially fail to generalize to real-world out-of-distribution scenarios . | As a result, these models could potentially fail to generalize to real-world out-of-distribution data . | clarity | 2103.06922 | 1 |
In this work, we show that the shortcut learning behavior can be explained by the long-tailed phenomenon . | <clarity> In this work, we show that the shortcut learning behavior can be explained by the long-tailed phenomenon . | In this work, we show that the words in the NLU training set can be modeled as a long-tailed phenomenon . | clarity | 2103.06922 | 1 |
In this work, we show that the shortcut learning behavior can be explained by the long-tailed phenomenon . | <clarity> In this work, we show that the shortcut learning behavior can be explained by the long-tailed phenomenon . | In this work, we show that the shortcut learning behavior can be explained by the long-tailed distribution . | clarity | 2103.06922 | 1 |
There are two findings: 1) Trained NLU models have strong preference for features located at the head of the long-tailed distribution, and 2) Shortcut features are picked up during very early few iterations of the model training. | <clarity> There are two findings: 1) Trained NLU models have strong preference for features located at the head of the long-tailed distribution, and 2) Shortcut features are picked up during very early few iterations of the model training. | There are two findings: 1) NLU models have strong preference for features located at the head of the long-tailed distribution, and 2) Shortcut features are picked up during very early few iterations of the model training. | clarity | 2103.06922 | 1 |
Based on this shortcut measurement, we propose a shortcut mitigation framework , to suppress the model from making overconfident predictions for samples with large shortcut degree. | <meaning-changed> Based on this shortcut measurement, we propose a shortcut mitigation framework , to suppress the model from making overconfident predictions for samples with large shortcut degree. | Based on this shortcut measurement, we propose a shortcut mitigation framework LGTR , to suppress the model from making overconfident predictions for samples with large shortcut degree. | meaning-changed | 2103.06922 | 1 |
Experimental analysis further indicates that our method can improve the generalization accuracy on OOD data, while preserving the accuracy on in distribution test data. | <clarity> Experimental analysis further indicates that our method can improve the generalization accuracy on OOD data, while preserving the accuracy on in distribution test data. | Experimental analysis further indicates that LGTR can improve the generalization accuracy on OOD data, while preserving the accuracy on in distribution test data. | clarity | 2103.06922 | 1 |
Experimental analysis further indicates that our method can improve the generalization accuracy on OOD data, while preserving the accuracy on in distribution test data. | <clarity> Experimental analysis further indicates that our method can improve the generalization accuracy on OOD data, while preserving the accuracy on in distribution test data. | Experimental analysis further indicates that our method can improve the generalization accuracy on OOD data, while preserving the accuracy on in-distribution data. | clarity | 2103.06922 | 1 |
Previous works on expressive text-to-speech (TTS) have a limitation on robustness and speed when training and inferring. | <meaning-changed> Previous works on expressive text-to-speech (TTS) have a limitation on robustness and speed when training and inferring. | Previous works on neural text-to-speech (TTS) have a limitation on robustness and speed when training and inferring. | meaning-changed | 2103.09474 | 1 |
Previous works on expressive text-to-speech (TTS) have a limitation on robustness and speed when training and inferring. Such drawbacks mostly come from autoregressive decoding, which makes the succeeding step vulnerable to preceding error. To overcome this weakness , we propose STYLER, a novel expressive text-to-speech model with parallelized architecture. | <clarity> Previous works on expressive text-to-speech (TTS) have a limitation on robustness and speed when training and inferring. Such drawbacks mostly come from autoregressive decoding, which makes the succeeding step vulnerable to preceding error. To overcome this weakness , we propose STYLER, a novel expressive text-to-speech model with parallelized architecture. | Previous works on expressive text-to-speech (TTS) have been tackled on limited speed in training and inference time, robustness for difficult synthesis conditions, expressiveness, and controllability. Although several approaches resolve some limitations, none of them has resolved all weaknesses at once. In this paper , we propose STYLER, a novel expressive text-to-speech model with parallelized architecture. | clarity | 2103.09474 | 1 |
To overcome this weakness , we propose STYLER, a novel expressive text-to-speech model with parallelized architecture. | <clarity> To overcome this weakness , we propose STYLER, a novel expressive text-to-speech model with parallelized architecture. | To overcome this weakness , we propose STYLER, an expressive and controllable text-to-speech model with parallelized architecture. | clarity | 2103.09474 | 1 |
To overcome this weakness , we propose STYLER, a novel expressive text-to-speech model with parallelized architecture. Expelling autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust even with high style transfer performance . | <clarity> To overcome this weakness , we propose STYLER, a novel expressive text-to-speech model with parallelized architecture. Expelling autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust even with high style transfer performance . | To overcome this weakness , we propose STYLER, a novel expressive text-to-speech model with robust speech synthesis and high speed. Excluding autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust even with high style transfer performance . | clarity | 2103.09474 | 1 |
Expelling autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust even with high style transfer performance . | <meaning-changed> Expelling autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust even with high style transfer performance . | Expelling autoregressive decoding and introducing a novel audio-text aligning method called Mel Calibrator leads speech synthesis more robust even with high style transfer performance . | meaning-changed | 2103.09474 | 1 |
Expelling autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust even with high style transfer performance . | <clarity> Expelling autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust even with high style transfer performance . | Expelling autoregressive decoding and introducing speech decomposition for encoding enables speech synthesis more robust on long, unseen data. Disentangled style factor modeling under supervision enlarges the controllability of synthesizing speech with fruitful expressivity . | clarity | 2103.09474 | 1 |
Moreover, our novel noise modeling approach from audio using domain adversarial training and Residual Decoding enabled style transferwithout transferring noise . | <clarity> Moreover, our novel noise modeling approach from audio using domain adversarial training and Residual Decoding enabled style transferwithout transferring noise . | Moreover, our novel noise modeling pipeline using domain adversarial training and Residual Decoding enabled style transferwithout transferring noise . | clarity | 2103.09474 | 1 |
Moreover, our novel noise modeling approach from audio using domain adversarial training and Residual Decoding enabled style transferwithout transferring noise . Our experiments prove the naturalness and expressiveness of our model from comparison with other parallel TTS models . Together we investigate our model 's robustness and speed by comparison with the expressive TTS model with autoregressive decoding . | <meaning-changed> Moreover, our novel noise modeling approach from audio using domain adversarial training and Residual Decoding enabled style transferwithout transferring noise . Our experiments prove the naturalness and expressiveness of our model from comparison with other parallel TTS models . Together we investigate our model 's robustness and speed by comparison with the expressive TTS model with autoregressive decoding . | Moreover, our novel noise modeling approach from audio using domain adversarial training and Residual Decoding enables noise-robust style transfer, decomposing the noise without any additional label. Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed, robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies. Synthesis samples of our model and experiment results are provided via our demo page . | meaning-changed | 2103.09474 | 1 |
Previous works on neural text-to-speech (TTS) have been tackled on limited speed in training and inference time, robustness for difficult synthesis conditions, expressiveness, and controllability. | <clarity> Previous works on neural text-to-speech (TTS) have been tackled on limited speed in training and inference time, robustness for difficult synthesis conditions, expressiveness, and controllability. | Previous works on neural text-to-speech (TTS) have been addressed on limited speed in training and inference time, robustness for difficult synthesis conditions, expressiveness, and controllability. | clarity | 2103.09474 | 2 |
Although several approaches resolve some limitations, none of them has resolved all weaknesses at once. | <clarity> Although several approaches resolve some limitations, none of them has resolved all weaknesses at once. | Although several approaches resolve some limitations, there has been no attempt to solve all weaknesses at once. | clarity | 2103.09474 | 2 |
In this paper, we propose STYLER, an expressive and controllable text-to-speech model with robust speech synthesisand high speed. Excluding autoregressive decoding and introducing a novel audio-text aligning method called Mel Calibrator leads speech synthesis more robust on long, unseen data. | <clarity> In this paper, we propose STYLER, an expressive and controllable text-to-speech model with robust speech synthesisand high speed. Excluding autoregressive decoding and introducing a novel audio-text aligning method called Mel Calibrator leads speech synthesis more robust on long, unseen data. | In this paper, we propose STYLER, an expressive and controllable TTS framework with high-speed and robust synthesis. Our novel audio-text aligning method called Mel Calibrator leads speech synthesis more robust on long, unseen data. | clarity | 2103.09474 | 2 |
Excluding autoregressive decoding and introducing a novel audio-text aligning method called Mel Calibrator leads speech synthesis more robust on long, unseen data. | <clarity> Excluding autoregressive decoding and introducing a novel audio-text aligning method called Mel Calibrator leads speech synthesis more robust on long, unseen data. | Excluding autoregressive decoding and introducing a novel audio-text aligning method called Mel Calibrator and excluding autoregressive decoding enable rapid training and inference and robust synthesis on unseen data. | clarity | 2103.09474 | 2 |
Disentangled style factor modeling under supervision enlarges the controllability of synthesizing speech with fruitful expressivity. | <coherence> Disentangled style factor modeling under supervision enlarges the controllability of synthesizing speech with fruitful expressivity. | Also, disentangled style factor modeling under supervision enlarges the controllability of synthesizing speech with fruitful expressivity. | coherence | 2103.09474 | 2 |
Disentangled style factor modeling under supervision enlarges the controllability of synthesizing speech with fruitful expressivity. Moreover, our novel noise modeling pipeline using domain adversarial training and Residual Decoding enables noise-robust style transfer, decomposing the noise without any additional label. | <clarity> Disentangled style factor modeling under supervision enlarges the controllability of synthesizing speech with fruitful expressivity. Moreover, our novel noise modeling pipeline using domain adversarial training and Residual Decoding enables noise-robust style transfer, decomposing the noise without any additional label. | Disentangled style factor modeling under supervision enlarges the controllability in synthesizing process leading to expressive TTS. On top of it, a novel noise modeling pipeline using domain adversarial training and Residual Decoding enables noise-robust style transfer, decomposing the noise without any additional label. | clarity | 2103.09474 | 2 |
Moreover, our novel noise modeling pipeline using domain adversarial training and Residual Decoding enables noise-robust style transfer, decomposing the noise without any additional label. | <clarity> Moreover, our novel noise modeling pipeline using domain adversarial training and Residual Decoding enables noise-robust style transfer, decomposing the noise without any additional label. | Moreover, our novel noise modeling pipeline using domain adversarial training and Residual Decoding empowers noise-robust style transfer, decomposing the noise without any additional label. | clarity | 2103.09474 | 2 |
Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples of our model and experiment results are provided via our demo page . | <clarity> Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples of our model and experiment results are provided via our demo page . | Various experiments demonstrate that STYLER is more effective in speed and robustness than expressive TTS with autoregressive decoding and more expressive and controllable than reading style non-autoregressive TTS . Synthesis samples of our model and experiment results are provided via our demo page . | clarity | 2103.09474 | 2 |
Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples of our model and experiment results are provided via our demo page . | <clarity> Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples of our model and experiment results are provided via our demo page . | Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples and experiment results are provided via our demo page . | clarity | 2103.09474 | 2 |
Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples of our model and experiment results are provided via our demo page . | <meaning-changed> Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples of our model and experiment results are provided via our demo page . | Our extensive and various experiments demonstrate STYLER's effectiveness in the aspects of speed , robustness, expressiveness, and controllability by comparison with existing neural TTS models and ablation studies . Synthesis samples of our model and experiment results are provided via our demo page , and code is available publicly . | meaning-changed | 2103.09474 | 2 |
Also , offenses and hate speech may occur through sarcasm, which hides the real intention of the comment and makes the decision of the annotators more confusing. | <coherence> Also , offenses and hate speech may occur through sarcasm, which hides the real intention of the comment and makes the decision of the annotators more confusing. | Moreover , offenses and hate speech may occur through sarcasm, which hides the real intention of the comment and makes the decision of the annotators more confusing. | coherence | 2103.14972 | 4 |
Therefore, provide a well-structured annotation process is crucial to a better understanding of hate speech and offensive language phenomena, as well as supply better performance for machine learning classifiers. | <fluency> Therefore, provide a well-structured annotation process is crucial to a better understanding of hate speech and offensive language phenomena, as well as supply better performance for machine learning classifiers. | Therefore, providing a well-structured annotation process is crucial to a better understanding of hate speech and offensive language phenomena, as well as supply better performance for machine learning classifiers. | fluency | 2103.14972 | 4 |
Therefore, provide a well-structured annotation process is crucial to a better understanding of hate speech and offensive language phenomena, as well as supply better performance for machine learning classifiers. | <fluency> Therefore, provide a well-structured annotation process is crucial to a better understanding of hate speech and offensive language phenomena, as well as supply better performance for machine learning classifiers. | Therefore, provide a well-structured annotation process is crucial to a better understanding of hate speech and offensive language phenomena, as well as supplying better performance for machine learning classifiers. | fluency | 2103.14972 | 4 |
In this paper, we describe a corpus annotation process , which was guided by a linguist, and a hate speech skilled to support the identification of hate speech and offensive language on social media. | <clarity> In this paper, we describe a corpus annotation process , which was guided by a linguist, and a hate speech skilled to support the identification of hate speech and offensive language on social media. | In this paper, we describe a corpus annotation process proposed by a linguist, and a hate speech skilled to support the identification of hate speech and offensive language on social media. | clarity | 2103.14972 | 4 |
In this paper, we describe a corpus annotation process , which was guided by a linguist, and a hate speech skilled to support the identification of hate speech and offensive language on social media. | <fluency> In this paper, we describe a corpus annotation process , which was guided by a linguist, and a hate speech skilled to support the identification of hate speech and offensive language on social media. | In this paper, we describe a corpus annotation process , which was guided by a linguist, a hate speech skilled to support the identification of hate speech and offensive language on social media. | fluency | 2103.14972 | 4 |
In this paper, we describe a corpus annotation process , which was guided by a linguist, and a hate speech skilled to support the identification of hate speech and offensive language on social media. | <meaning-changed> In this paper, we describe a corpus annotation process , which was guided by a linguist, and a hate speech skilled to support the identification of hate speech and offensive language on social media. | In this paper, we describe a corpus annotation process , which was guided by a linguist, and a hate speech specialist, and machine learning engineers in order to support the identification of hate speech and offensive language on social media. | meaning-changed | 2103.14972 | 4 |
In addition, we provide the first robust corpus of this kind for the Brazilian Portuguese language. | <clarity> In addition, we provide the first robust corpus of this kind for the Brazilian Portuguese language. | In addition, we provide the first robust dataset of this kind for the Brazilian Portuguese language. | clarity | 2103.14972 | 4 |
The corpus was collected from Instagram posts of political personalities and manually annotated, being composed by 7,000 documents annotated according to three different layers: | <fluency> The corpus was collected from Instagram posts of political personalities and manually annotated, being composed by 7,000 documents annotated according to three different layers: | The corpus was collected from Instagram posts of political personalities and manually annotated, being composed by 7,000 annotated documents according to three different layers: | fluency | 2103.14972 | 4 |
a binary classification (offensive versus non-offensive language), the level of offense (highly offensive, moderately offensive, and slightly offensive messages), and the identification regarding the target of the discriminatory content (xenophobia, racism, homophobia, sexism, religious intolerance, partyism, apology to the dictatorship, antisemitism, and fat phobia ). | <fluency> a binary classification (offensive versus non-offensive language), the level of offense (highly offensive, moderately offensive, and slightly offensive messages), and the identification regarding the target of the discriminatory content (xenophobia, racism, homophobia, sexism, religious intolerance, partyism, apology to the dictatorship, antisemitism, and fat phobia ). | a binary classification (offensive versus non-offensive language), the level of offense (highly offensive, moderately offensive, and slightly offensive messages), and the identification regarding the target of the discriminatory content (xenophobia, racism, homophobia, sexism, religious intolerance, partyism, apology to the dictatorship, antisemitism, and fatphobia ). | fluency | 2103.14972 | 4 |
Each comment was annotated by three different annotators , and achieved high inter-annotator agreement. | <fluency> Each comment was annotated by three different annotators , and achieved high inter-annotator agreement. | Each comment was annotated by three different annotators and achieved high inter-annotator agreement. | fluency | 2103.14972 | 4 |
The new proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese ) . | <coherence> The new proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese ) . | The proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese ) . | coherence | 2103.14972 | 4 |
The new proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese ) . | <coherence> The new proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese ) . | The new proposed annotation approach is also language and domain-independent nevertheless it is currently customized for Brazilian Portuguese ) . | coherence | 2103.14972 | 4 |
The new proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese ) . | <fluency> The new proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese ) . | The new proposed annotation approach is also language and domain-independent (although it has been applied for Brazilian Portuguese . | fluency | 2103.14972 | 4 |
This paper presents a new approach for offensive language and hate speech detection on social media. | <clarity> This paper presents a new approach for offensive language and hate speech detection on social media. | This paper provides a new approach for offensive language and hate speech detection on social media. | clarity | 2104.12265 | 2 |
Our approach incorporate an offensive lexicon composed by implicit and explicit offensive and swearing expressions annotated with binary classes: context-dependent and context-independent offensive. | <fluency> Our approach incorporate an offensive lexicon composed by implicit and explicit offensive and swearing expressions annotated with binary classes: context-dependent and context-independent offensive. | Our approach incorporates an offensive lexicon composed by implicit and explicit offensive and swearing expressions annotated with binary classes: context-dependent and context-independent offensive. | fluency | 2104.12265 | 2 |
Our approach incorporate an offensive lexicon composed by implicit and explicit offensive and swearing expressions annotated with binary classes: context-dependent and context-independent offensive. | <fluency> Our approach incorporate an offensive lexicon composed by implicit and explicit offensive and swearing expressions annotated with binary classes: context-dependent and context-independent offensive. | Our approach incorporate an offensive lexicon composed of implicit and explicit offensive and swearing expressions annotated with binary classes: context-dependent and context-independent offensive. | fluency | 2104.12265 | 2 |
Due to the severity of the hate speech and offensive comments in Brazil, and the lack of research in Portuguese, Brazilian Portuguese is the language used to validate our method. However, the proposal may be applied to any other language or domain. | <coherence> Due to the severity of the hate speech and offensive comments in Brazil, and the lack of research in Portuguese, Brazilian Portuguese is the language used to validate our method. However, the proposal may be applied to any other language or domain. | Due to the severity of the hate speech and offensive comments in Brazil, and the lack of research in Portuguese, Brazilian Portuguese is the language used to validate the proposed method. Nevertheless, our proposal may be applied to any other language or domain. | coherence | 2104.12265 | 2 |
Based on the obtained results, the proposed approach showed high performance results overcoming the current baselines for European and Brazilian Portuguese. | <fluency> Based on the obtained results, the proposed approach showed high performance results overcoming the current baselines for European and Brazilian Portuguese. | Based on the obtained results, the proposed approach showed high-performance overcoming the current baselines for European and Brazilian Portuguese. | fluency | 2104.12265 | 2 |
In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension ( ViCoQA ), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | <clarity> In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension ( ViCoQA ), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension ( UIT-ViCoQA ), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | clarity | 2105.01542 | 1 |
We analyze ViCoQA in depth with different linguistic aspects. | <clarity> We analyze ViCoQA in depth with different linguistic aspects. | We analyze UIT-ViCoQA in depth with different linguistic aspects. | clarity | 2105.01542 | 1 |
Then, we evaluate several baseline models about dialogue and reading comprehension on the ViCoQA corpus. | <clarity> Then, we evaluate several baseline models about dialogue and reading comprehension on the ViCoQA corpus. | Then, we evaluate several baseline models about dialogue and reading comprehension on the UIT-ViCoQA corpus. | clarity | 2105.01542 | 1 |
Machine reading comprehension (MRC) is a sub-field in natural language processing or computational linguistics. MRC aims to help computers understand unstructured texts and then answer questions related to them. | <clarity> Machine reading comprehension (MRC) is a sub-field in natural language processing or computational linguistics. MRC aims to help computers understand unstructured texts and then answer questions related to them. | Machine reading comprehension (MRC) is a sub-field in natural language processing which aims to help computers understand unstructured texts and then answer questions related to them. | clarity | 2105.01542 | 2 |
In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | <coherence> In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | In practice, conversation is an essential way to communicate and transfer information. To help machines understand conversation texts, we present UIT-ViCoQA - a new corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | coherence | 2105.01542 | 2 |
In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | <meaning-changed> In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension in the Vietnamese language. This corpus consists of 10,000 questions with answers over 2,000 conversations about health news articles. | meaning-changed | 2105.01542 | 2 |
In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | <fluency> In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers over 2,000 conversations about health news articles. | In this paper, we present a new Vietnamese corpus for conversational machine reading comprehension (UIT-ViCoQA), consisting of 10,000 questions with answers to over 2,000 conversations about health news articles. | fluency | 2105.01542 | 2 |
We analyze UIT-ViCoQA in depth with different linguistic aspects. | <coherence> We analyze UIT-ViCoQA in depth with different linguistic aspects. | coherence | 2105.01542 | 2 |
|
Then, we evaluate several baseline models about dialogue and reading comprehension on the UIT-ViCoQA corpus. | <clarity> Then, we evaluate several baseline models about dialogue and reading comprehension on the UIT-ViCoQA corpus. | Then, we evaluate several baseline approaches for conversational machine comprehension on the UIT-ViCoQA corpus. | clarity | 2105.01542 | 2 |
During a meeting of the Democratic Party of Serbia today , the Prime Minister Vojislav Kotunica said that he will not agree to Kosovo's independence and that the solution for Kosovo is wide autonomy. | <meaning-changed> During a meeting of the Democratic Party of Serbia today , the Prime Minister Vojislav Kotunica said that he will not agree to Kosovo's independence and that the solution for Kosovo is wide autonomy. | During a meeting of the Democratic Party of Serbia on Saturday , the Prime Minister Vojislav Kotunica said that he will not agree to Kosovo's independence and that the solution for Kosovo is wide autonomy. | meaning-changed | 10010 | 1 |
During a meeting of the Democratic Party of Serbia today , the Prime Minister Vojislav Kotunica said that he will not agree to Kosovo's independence and that the solution for Kosovo is wide autonomy. | <fluency> During a meeting of the Democratic Party of Serbia today , the Prime Minister Vojislav Kotunica said that he will not agree to Kosovo's independence and that the solution for Kosovo is wide autonomy. | During a meeting of the Democratic Party of Serbia today , the Prime Minister Vojislav Kotunica said that he will not agree to Kosovo's independence , and that the solution for Kosovo is wide autonomy. | fluency | 10010 | 1 |
After the war in 1999, Kosovo was put under UNMIK's administration, but it still officialy remains a province in Serbia. | <fluency> After the war in 1999, Kosovo was put under UNMIK's administration, but it still officialy remains a province in Serbia. | After the war in 1999, Kosovo was put under UNMIK's administration, but it still officially remains a province in Serbia. | fluency | 10010 | 1 |
He also reaffirmed his stance on participation of Serbs in Kosovo's institutions: "For now, there is no room for Serbian representatives in institutions of Kosovo." | <clarity> He also reaffirmed his stance on participation of Serbs in Kosovo's institutions: "For now, there is no room for Serbian representatives in institutions of Kosovo." | Kotunica also reaffirmed his stance on participation of Serbs in Kosovo's institutions: "For now, there is no room for Serbian representatives in institutions of Kosovo." | clarity | 10010 | 1 |
The sites sued are Shuntv.net, Zonatracker.com, Btefnet.net, Scifi-classics.net, Cddvdheaven.co.uk, and Bragginrights.biz According to research by Envisional, recent surveys report downloads of TV programmes are up by 150\% in a year. | <fluency> The sites sued are Shuntv.net, Zonatracker.com, Btefnet.net, Scifi-classics.net, Cddvdheaven.co.uk, and Bragginrights.biz According to research by Envisional, recent surveys report downloads of TV programmes are up by 150\% in a year. | The sites sued are Shuntv.net, Zonatracker.com, Btefnet.net, Scifi-classics.net, Cddvdheaven.co.uk, and Bragginrights.biz . According to research by Envisional, recent surveys report downloads of TV programmes are up by 150\% in a year. | fluency | 10044 | 3 |
With smaller size than movies and broadband access, popular shows often appear within hours of airing on TV. | <clarity> With smaller size than movies and broadband access, popular shows often appear within hours of airing on TV. | With the increasing ubiquity of broadband Internet access, popular shows often appear within hours of airing on TV. | clarity | 10044 | 3 |
With smaller size than movies and broadband access, popular shows often appear within hours of airing on TV. | <meaning-changed> With smaller size than movies and broadband access, popular shows often appear within hours of airing on TV. | With smaller size than movies and broadband access, popular television shows, which are smaller downloads than movies, often appear within hours of airing on TV. | meaning-changed | 10044 | 3 |
CEO of the MPAA , Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | <clarity> CEO of the MPAA , Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | MPAA CEO Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | clarity | 10044 | 3 |
CEO of the MPAA , Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | <fluency> CEO of the MPAA , Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | CEO of the MPAA , Dan Glickman told the BBC , "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | fluency | 10044 | 3 |
CEO of the MPAA , Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | <fluency> CEO of the MPAA , Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially which means the experience of downloading copyrighted films and TV shows is not what it used to be. | CEO of the MPAA , Dan Glickman told the BBC "Since we began shutting these sites down, the time that it takes to download a file on BitTorrent has increased exponentially , which means the experience of downloading copyrighted films and TV shows is not what it used to be. | fluency | 10044 | 3 |
Elections for local governments were held in Croatia today . | <meaning-changed> Elections for local governments were held in Croatia today . | Elections for local governments were held in Croatia Sunday . | meaning-changed | 10050 | 1 |
According to GONG, an NGO observing the elections, Milan Bandić will be able to form a government in the city of Zagreb, since his list has won around 46\%. | <clarity> According to GONG, an NGO observing the elections, Milan Bandić will be able to form a government in the city of Zagreb, since his list has won around 46\%. | According to GONG, an NGO observing the elections, Milan Bandić , can form a government in the city of Zagreb, since his list has won around 46\%. | clarity | 10050 | 1 |
This would give them 27 of 51 seats in the capital city. | <clarity> This would give them 27 of 51 seats in the capital city. | This would give the coalition 27 of 51 seats in the capital city. | clarity | 10050 | 1 |
Branimir Glavaš's, who was recently expelled from Croatian Democratic Union (HDZ), list seems to be in lead in Osijek according to the exit polls, with 27.24\% (which would equal 9 seats in the local parliament) , Liberal Party has 18.47\% (6 seats) and Croatian Party of Justice is third with 14.5\% (4 seats). | <fluency> Branimir Glavaš's, who was recently expelled from Croatian Democratic Union (HDZ), list seems to be in lead in Osijek according to the exit polls, with 27.24\% (which would equal 9 seats in the local parliament) , Liberal Party has 18.47\% (6 seats) and Croatian Party of Justice is third with 14.5\% (4 seats). | Branimir Glavaš's, who was recently expelled from Croatian Democratic Union (HDZ), list seems to be in lead in Osijek according to the exit polls, with 27.24\% (which would equal 9 seats in the local parliament) . Liberal Party has 18.47\% (6 seats) and Croatian Party of Justice is third with 14.5\% (4 seats). | fluency | 10050 | 1 |
Branimir Glavaš's, who was recently expelled from Croatian Democratic Union (HDZ), list seems to be in lead in Osijek according to the exit polls, with 27.24\% (which would equal 9 seats in the local parliament) , Liberal Party has 18.47\% (6 seats) and Croatian Party of Justice is third with 14.5\% (4 seats). | <fluency> Branimir Glavaš's, who was recently expelled from Croatian Democratic Union (HDZ), list seems to be in lead in Osijek according to the exit polls, with 27.24\% (which would equal 9 seats in the local parliament) , Liberal Party has 18.47\% (6 seats) and Croatian Party of Justice is third with 14.5\% (4 seats). | Branimir Glavaš's, who was recently expelled from Croatian Democratic Union (HDZ), list seems to be in lead in Osijek according to the exit polls, with 27.24\% (which would equal 9 seats in the local parliament) , Liberal Party has 18.47\% (6 seats) , and Croatian Party of Justice is third with 14.5\% (4 seats). | fluency | 10050 | 1 |
According to GONG, Social Democrat Party has a lead in Split with 29.02\% of votes (10 seats), while Croatian Democratic Union is following with 18.54\% (6 seats). | <fluency> According to GONG, Social Democrat Party has a lead in Split with 29.02\% of votes (10 seats), while Croatian Democratic Union is following with 18.54\% (6 seats). | According to GONG, Social Democrat Party has a lead in split with 29.02\% of votes (10 seats), while Croatian Democratic Union is following with 18.54\% (6 seats). | fluency | 10050 | 1 |
The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at Taipei Howard Plaza Hotel ; | <fluency> The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at Taipei Howard Plaza Hotel ; | The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) returned to Taiwan this year at Taipei Howard Plaza Hotel ; | fluency | 100524 | 1 |
The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at Taipei Howard Plaza Hotel ; | <clarity> The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at Taipei Howard Plaza Hotel ; | The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at the Taipei Howard Plaza Hotel ; | clarity | 100524 | 1 |
The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at Taipei Howard Plaza Hotel ; | <coherence> The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at Taipei Howard Plaza Hotel ; | The twelfth-annual Asia Pacific Regional Internet Conference on Operational Technologies (a.k.a APRICOT) , returned to Taiwan this year at Taipei Howard Plaza Hotel . This is | coherence | 100524 | 1 |
As Wikinews Journalist Rico Shen reported on the recent "Edison Chen photo scandal" incident, he commented: Workshops with varied topics and different technology levels took place from February 20 to 24, while several main seminars and speeches for industry, governmental, and academic executives ran from February 25 to 29. | <clarity> As Wikinews Journalist Rico Shen reported on the recent "Edison Chen photo scandal" incident, he commented: Workshops with varied topics and different technology levels took place from February 20 to 24, while several main seminars and speeches for industry, governmental, and academic executives ran from February 25 to 29. | When Wikinews journalist Rico Shen reported on the recent "Edison Chen photo scandal" incident, he commented: Workshops with varied topics and different technology levels took place from February 20 to 24, while several main seminars and speeches for industry, governmental, and academic executives ran from February 25 to 29. | clarity | 100524 | 1 |
As Wikinews Journalist Rico Shen reported on the recent "Edison Chen photo scandal" incident, he commented: Workshops with varied topics and different technology levels took place from February 20 to 24, while several main seminars and speeches for industry, governmental, and academic executives ran from February 25 to 29. | <meaning-changed> As Wikinews Journalist Rico Shen reported on the recent "Edison Chen photo scandal" incident, he commented: Workshops with varied topics and different technology levels took place from February 20 to 24, while several main seminars and speeches for industry, governmental, and academic executives ran from February 25 to 29. | As Wikinews Journalist Rico Shen reported on the recent "Edison Chen photo scandal" incident, Tony Tien-lai Teng commented: Workshops with varied topics and different technology levels took place from February 20 to 24, while several main seminars and speeches for industry, governmental, and academic executives ran from February 25 to 29. | meaning-changed | 100524 | 1 |
Several industry experts such as Wilfred Kwan (Chief Technology Officer of AsiaNetCom), Chung-laung Liu ( ISOC Taiwan Chapter Chair ), and Maemura Akinori (EC Chair of APNIC) will give several speeches related to the Internet industry at the conference. | <clarity> Several industry experts such as Wilfred Kwan (Chief Technology Officer of AsiaNetCom), Chung-laung Liu ( ISOC Taiwan Chapter Chair ), and Maemura Akinori (EC Chair of APNIC) will give several speeches related to the Internet industry at the conference. | Several industry experts such as Wilfred Kwan (Chief Technology Officer of AsiaNetCom), Chung-laung Liu ( Taiwan Chapter Chair ), and Maemura Akinori (EC Chair of APNIC) will give several speeches related to the Internet industry at the conference. | clarity | 100524 | 1 |
Several industry experts such as Wilfred Kwan (Chief Technology Officer of AsiaNetCom), Chung-laung Liu ( ISOC Taiwan Chapter Chair ), and Maemura Akinori (EC Chair of APNIC) will give several speeches related to the Internet industry at the conference. | <meaning-changed> Several industry experts such as Wilfred Kwan (Chief Technology Officer of AsiaNetCom), Chung-laung Liu ( ISOC Taiwan Chapter Chair ), and Maemura Akinori (EC Chair of APNIC) will give several speeches related to the Internet industry at the conference. | Several industry experts such as Wilfred Kwan (Chief Technology Officer of AsiaNetCom), Chung-laung Liu ( ISOC Taiwan Chapter Chair of ISOC ), and Maemura Akinori (EC Chair of APNIC) will give several speeches related to the Internet industry at the conference. | meaning-changed | 100524 | 1 |
ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | <meaning-changed> ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | ZipcodeZoo offers descriptions of over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | meaning-changed | 100534 | 1 |
ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | <clarity> ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | ZipcodeZoo offers over 3 million species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | clarity | 100534 | 1 |
ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | <coherence> ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. | ZipcodeZoo offers over 3 million web pages describing species of plants and animals , 888 terms. | coherence | 100534 | 1 |
ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. 70 , 725 Large photos can be zoomed and panned . ARKive collects media of species, and the All Species Foundation was a failed early attempt at a web catalog. | <meaning-changed> ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , 888 terms. 70 , 725 Large photos can be zoomed and panned . ARKive collects media of species, and the All Species Foundation was a failed early attempt at a web catalog. | ZipcodeZoo offers over 3 million web pages describing species of plants and animals . Pages contain 258, 753 photostaken by 1, 369 photographers, 1, 104 sound recordings, and definitions of 234 , with 250 , 725 Large photos can be zoomed and panned . ARKive collects media of species, and the All Species Foundation was a failed early attempt at a web catalog. | meaning-changed | 100534 | 1 |
70 , 725 Large photos can be zoomed and panned . ARKive collects media of species, and the All Species Foundation was a failed early attempt at a web catalog. | <meaning-changed> 70 , 725 Large photos can be zoomed and panned . ARKive collects media of species, and the All Species Foundation was a failed early attempt at a web catalog. | 70 , 000 photos, maps, pronunciations, and definitions in 13 languages . ARKive collects media of species, and the All Species Foundation was a failed early attempt at a web catalog. | meaning-changed | 100534 | 1 |
Oil in productionCrude oil prices in New York rose to a new record of $102.59 per barrel on Thursday, although the figure increased even more during after hours trading. | <clarity> Oil in productionCrude oil prices in New York rose to a new record of $102.59 per barrel on Thursday, although the figure increased even more during after hours trading. | An oil refineryCrude oil prices in New York rose to a new record of $102.59 per barrel on Thursday, although the figure increased even more during after hours trading. | clarity | 100809 | 1 |
Oil in productionCrude oil prices in New York rose to a new record of $102.59 per barrel on Thursday, although the figure increased even more during after hours trading. | <meaning-changed> Oil in productionCrude oil prices in New York rose to a new record of $102.59 per barrel on Thursday, although the figure increased even more during after hours trading. | Oil in productionCrude oil prices in New York rose to a new record of US $102.59 per barrel on Thursday, although the figure increased even more during after hours trading. | meaning-changed | 100809 | 1 |
In less than a month, prices have risen $10, leading the figures to above the record highs set during the 1980s (taking inflation into account) . | <clarity> In less than a month, prices have risen $10, leading the figures to above the record highs set during the 1980s (taking inflation into account) . | In less than a month, prices have risen $10, leading the inflation adjusted prices above the record highs set during the 1980s (taking inflation into account) . | clarity | 100809 | 1 |
In less than a month, prices have risen $10, leading the figures to above the record highs set during the 1980s (taking inflation into account) . | <clarity> In less than a month, prices have risen $10, leading the figures to above the record highs set during the 1980s (taking inflation into account) . | In less than a month, prices have risen $10, leading the figures to above the record highs set during the 1980s . | clarity | 100809 | 1 |
The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | <clarity> The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | While the weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | clarity | 100809 | 1 |
The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | <fluency> The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | The weak dollar is seen as a major cause of this rise , Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | fluency | 100809 | 1 |
The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | <meaning-changed> The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the rapid ascent the of price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | meaning-changed | 100809 | 1 |
The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | <clarity> The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil 's rapid ascent , it had remained flat when compared to the price of gold. | The weak dollar is seen as a major cause of this rise . Congressman Ron Paul of Texas, pointed out to Federal Reserve chairman Ben Bernanke in a committee meeting this week that despite the price of oil , it had remained flat when compared to the price of gold. | clarity | 100809 | 1 |
Increasing demand for oil has also been cited as the cause for this increase. | <meaning-changed> Increasing demand for oil has also been cited as the cause for this increase. | A graph showing recent increases in gasoline prices Increasing demand for oil has also been cited as the cause for this increase. | meaning-changed | 100809 | 1 |
A graph to show the increase in gasoline pricesThere have also been suggestions that reports of a fire at a National Gas Terminal may have contributed to the rising oil price. | <clarity> A graph to show the increase in gasoline pricesThere have also been suggestions that reports of a fire at a National Gas Terminal may have contributed to the rising oil price. | There have also been suggestions that reports of a fire at a National Gas Terminal may have contributed to the rising oil price. | clarity | 100809 | 1 |
Time Evans from Citigroup Futures has stated he believes that this fire at the UK natural gas terminal is creating a strong push in the European market, and that is translating here [the US]." | <clarity> Time Evans from Citigroup Futures has stated he believes that this fire at the UK natural gas terminal is creating a strong push in the European market, and that is translating here [the US]." | Time Evans from Citigroup Futures said he believes " that this fire at the UK natural gas terminal is creating a strong push in the European market, and that is translating here [the US]." | clarity | 100809 | 1 |
The H5N1 Avian Flu Virus has been found in a dead wild Canadian Goose in Abbotsbury Swannery in Dorset, England. | <fluency> The H5N1 Avian Flu Virus has been found in a dead wild Canadian Goose in Abbotsbury Swannery in Dorset, England. | The H5N1 Avian Flu virus has been found in a dead wild Canadian Goose in Abbotsbury Swannery in Dorset, England. | fluency | 100868 | 1 |
This is the 11th case of the virus turning up in wild birds. | <fluency> This is the 11th case of the virus turning up in wild birds. | This is the eleventh case of the virus turning up in wild birds. | fluency | 100868 | 1 |
The goose was found on February 25, 2008. | <style> The goose was found on February 25, 2008. | The goose was discovered on February 25, 2008. | style | 100868 | 1 |
10 other cases of the virus have appeared in dead birds, all Mute Swans from the same area. | <clarity> 10 other cases of the virus have appeared in dead birds, all Mute Swans from the same area. | Ten other cases of the virus have appeared in dead birds, all Mute Swans from the same area. | clarity | 100868 | 1 |
Interpol has released an "Orange Notice" for the Southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | <clarity> Interpol has released an "Orange Notice" for the Southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | Interpol has issued an "Orange Notice" for the Southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | clarity | 100873 | 1 |
Interpol has released an "Orange Notice" for the Southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | <clarity> Interpol has released an "Orange Notice" for the Southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | Interpol has released an "Orange Notice" for the leader of southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | clarity | 100873 | 1 |
Interpol has released an "Orange Notice" for the Southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | <coherence> Interpol has released an "Orange Notice" for the Southeast Asian Jemaah Islamiyah terrorist group leader Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | Interpol has released an "Orange Notice" for the Southeast Asian Jemaah Islamiyah , Mas Selamat bin Kastari, who escaped from a detention center Wednesday. | coherence | 100873 | 1 |