Pub Date : 2009-09-01DOI: 10.30019/IJCLCLP.200909.0001
Un-Gian Iunn, Jia-hung Tai, K. Lau, Cheng-Yan Kao, Keh-Jiann Chen
In this paper, we introduce a POS tagging method for Taiwan Southern Min. We use the more than 62,000 entries of the Taiwanese-Mandarin dictionary and 10 million words of Mandarin training data to tag Taiwanese. The literary written Taiwanese corpora have both Romanized script and Han-Romanization mixed script, and include prose, novels, and dramas. We follow the tagset drawn up by CKIP. We developed a word alignment checker to assist with the word alignment for the two scripts. It searches the Taiwanese-Mandarin dictionary to find corresponding Mandarin candidate words, selects the most suitable Mandarin word using an HMM probabilistic model from the Mandarin training data, and tags the word using an MEMM classifier. We achieve an accuracy rate of 91.6% on Taiwanese POS tagging work, and we analyze the errors. We also discover some preliminary Taiwanese training data.
{"title":"Modeling Taiwanese POS Tagging Using Statistical Methods and Mandarin Training Data","authors":"Un-Gian Iunn, Jia-hung Tai, K. Lau, Cheng-Yan Kao, Keh-Jiann Chen","doi":"10.30019/IJCLCLP.200909.0001","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200909.0001","url":null,"abstract":"In this paper, we introduce a POS tagging method for Taiwan Southern Min. We use the more than 62,000 entries of the Taiwanese-Mandarin dictionary and 10 million words of Mandarin training data to tag Taiwanese. The literary written Taiwanese corpora have both Romanized script and Han-Romanization mixed script, and include prose, novels, and dramas. We follow the tagset drawn up by CKIP. We developed a word alignment checker to assist with the word alignment for the two scripts. It searches the Taiwanese-Mandarin dictionary to find corresponding Mandarin candidate words, selects the most suitable Mandarin word using an HMM probabilistic model from the Mandarin training data, and tags the word using an MEMM classifier. We achieve an accuracy rate of 91.6% on Taiwanese POS tagging work, and we analyze the errors. We also discover some preliminary Taiwanese training data.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124174636","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2009-09-01DOI: 10.30019/IJCLCLP.200909.0003
Joyce Y. C. Chan, Houwei Cao, P. Ching, Tan Lee
Code-mixing is a common phenomenon in bilingual societies. It refers to the intra-sentential switching of two different languages in a spoken utterance. This paper presents the first study on automatic recognition of Cantonese-English code-mixing speech, which is common in Hong Kong. This study starts with the design and compilation of code-mixing speech and text corpora. The problems of acoustic modeling, language modeling, and language boundary detection are investigated. Subsequently, a large-vocabulary code-mixing speech recognition system is developed based on a two-pass decoding algorithm. For acoustic modeling, it is shown that cross-lingual acoustic models are more appropriate than language-dependent models. The language models being used are character tri-grams, in which the embedded English words are grouped into a small number of classes. Language boundary detection is done either by exploiting the phonological and lexical differences between the two languages or is done based on the result of cross-lingual speech recognition. The language boundary information is used to re-score the hypothesized syllables or words in the decoding process. The proposed code-mixing speech recognition system attains the accuracies of 56.4% and 53.0% for the Cantonese syllables and English words in code-mixing utterances.
{"title":"Automatic Recognition of Cantonese-English Code-Mixing Speech","authors":"Joyce Y. C. Chan, Houwei Cao, P. Ching, Tan Lee","doi":"10.30019/IJCLCLP.200909.0003","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200909.0003","url":null,"abstract":"Code-mixing is a common phenomenon in bilingual societies. It refers to the intra-sentential switching of two different languages in a spoken utterance. This paper presents the first study on automatic recognition of Cantonese-English code-mixing speech, which is common in Hong Kong. This study starts with the design and compilation of code-mixing speech and text corpora. The problems of acoustic modeling, language modeling, and language boundary detection are investigated. Subsequently, a large-vocabulary code-mixing speech recognition system is developed based on a two-pass decoding algorithm. For acoustic modeling, it is shown that cross-lingual acoustic models are more appropriate than language-dependent models. The language models being used are character tri-grams, in which the embedded English words are grouped into a small number of classes. Language boundary detection is done either by exploiting the phonological and lexical differences between the two languages or is done based on the result of cross-lingual speech recognition. The language boundary information is used to re-score the hypothesized syllables or words in the decoding process. The proposed code-mixing speech recognition system attains the accuracies of 56.4% and 53.0% for the Cantonese syllables and English words in code-mixing utterances.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"63 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126347014","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2009-03-01DOI: 10.30019/IJCLCLP.200903.0001
Chung-Chi Huang, Jason J. S. Chang
We propose a version of Inversion Transduction Grammar (ITG) model with IBM-style notation of fertility to improve word-alignment performance. In our approach, binary context-free grammar rules of the source language, accompanied by orientation preferences of the target language and fertilities of words, are leveraged to construct a syntax-based statistical translation model. Our model, inherently possessing the characteristics of ITG restrictions and allowing for many consecutive words aligned to one and vice-versa, outperforms the Bracketing Transduction Grammar (BTG) model and GIZA++, a state-of-the-art word aligner, not only in alignment error rate (23% and 14% error reduction) but also in consistent phrase error rate (13% and 9% error reduction). Better performance in these two evaluation metrics suggests that, based on our word alignment result, more accurate phrase pairs may be acquired, leading to better machine translation quality.
{"title":"Fertility-based Source-Language-biased Inversion Transduction Grammar for Word Alignment","authors":"Chung-Chi Huang, Jason J. S. Chang","doi":"10.30019/IJCLCLP.200903.0001","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200903.0001","url":null,"abstract":"We propose a version of Inversion Transduction Grammar (ITG) model with IBM-style notation of fertility to improve word-alignment performance. In our approach, binary context-free grammar rules of the source language, accompanied by orientation preferences of the target language and fertilities of words, are leveraged to construct a syntax-based statistical translation model. Our model, inherently possessing the characteristics of ITG restrictions and allowing for many consecutive words aligned to one and vice-versa, outperforms the Bracketing Transduction Grammar (BTG) model and GIZA++, a state-of-the-art word aligner, not only in alignment error rate (23% and 14% error reduction) but also in consistent phrase error rate (13% and 9% error reduction). Better performance in these two evaluation metrics suggests that, based on our word alignment result, more accurate phrase pairs may be acquired, leading to better machine translation quality.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116213187","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2009-03-01DOI: 10.30019/IJCLCLP.200903.0003
Shu-Yen Lin, Cheng-chao Su, Yuda Lai, Li-Chin Yang, S. Hsieh
Although some traditional readability formulas have shown high predictive validity in the r=0.8 range and above (Chall & Dale, 1995), they are generally not based on genuine linguistic processing factors, but on statistical correlations (Crossley et al., 2008). Improvement of readability assessment should focus on finding variables that truly represent the comprehensibility of text as well as the indices that accurately measure the correlations. In this study, we explore the hierarchical relations between lexical items based on the conceptual categories advanced from Prototype Theory (Rosch et al., 1976). According to this theory and its development, basic level words like guitar represent the objects humans interact with most readily. They are acquired by children earlier than their superordinate words like stringed instrument and their subordinate words like acoustic guitar. Accordingly, the readability of a text is presumably associated with the ratio of basic level words it contains. WordNet (Fellbaum, 1998), a network of meaningfully related words, provides the best online open source database for studying such lexical relations. Our study shows that a basic level noun can be identified by its ratio of forming compounds (e.g. chair→armchair) and the length difference in relation to its hyponyms. We compared graded readings for American children and high school English readings for Taiwanese students by several readability formulas and in terms of basic level noun ratios (i.e. the number of basic level noun types divided by the number of noun types in a text). It is suggested that basic level noun ratios provide a robust and meaningful index of lexical complexity, which is directly associated with text readability.
虽然一些传统的可读性公式在r=0.8及以上的范围内显示出较高的预测效度(Chall & Dale, 1995)，但它们通常不是基于真正的语言处理因素，而是基于统计相关性(Crossley et al.， 2008)。可读性评价的改进应着眼于寻找真正代表文本可理解性的变量和准确衡量相关性的指标。在本研究中，我们在原型理论(Rosch et al.， 1976)提出的概念范畴的基础上，探讨了词汇项目之间的层次关系。根据这一理论及其发展，像吉他这样的基本词汇代表了人类最容易接触的对象。他们比他们的上级词如弦乐器和下级词如原声吉他更早被儿童习得。因此，文本的可读性大概与它所包含的基本级单词的比例有关。WordNet (Fellbaum, 1998)是一个有意义相关词汇的网络，为研究这种词汇关系提供了最好的在线开源数据库。我们的研究表明，基本级名词可以通过其构成复合词的比例(例如chair→armchair)和相对于其下位词的长度差异来识别。我们用几个可读性公式和基础水平名词比率(即基础水平名词类型的数量除以文本中名词类型的数量)来比较美国儿童和台湾学生的高中英语阅读。本文认为，基本水平名词比例为词汇复杂性提供了一个可靠而有意义的指标，而词汇复杂性与文本可读性直接相关。
{"title":"Assessing Text Readability Using Hierarchical Lexical Relations Retrieved from WordNet","authors":"Shu-Yen Lin, Cheng-chao Su, Yuda Lai, Li-Chin Yang, S. Hsieh","doi":"10.30019/IJCLCLP.200903.0003","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200903.0003","url":null,"abstract":"Although some traditional readability formulas have shown high predictive validity in the r=0.8 range and above (Chall & Dale, 1995), they are generally not based on genuine linguistic processing factors, but on statistical correlations (Crossley et al., 2008). Improvement of readability assessment should focus on finding variables that truly represent the comprehensibility of text as well as the indices that accurately measure the correlations. In this study, we explore the hierarchical relations between lexical items based on the conceptual categories advanced from Prototype Theory (Rosch et al., 1976). According to this theory and its development, basic level words like guitar represent the objects humans interact with most readily. They are acquired by children earlier than their superordinate words like stringed instrument and their subordinate words like acoustic guitar. Accordingly, the readability of a text is presumably associated with the ratio of basic level words it contains. WordNet (Fellbaum, 1998), a network of meaningfully related words, provides the best online open source database for studying such lexical relations. Our study shows that a basic level noun can be identified by its ratio of forming compounds (e.g. chair→armchair) and the length difference in relation to its hyponyms. We compared graded readings for American children and high school English readings for Taiwanese students by several readability formulas and in terms of basic level noun ratios (i.e. the number of basic level noun types divided by the number of noun types in a text). It is suggested that basic level noun ratios provide a robust and meaningful index of lexical complexity, which is directly associated with text readability.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2009-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128067627","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-12-01DOI: 10.30019/IJCLCLP.200812.0002
Liang-Chih Yu, Chung-Hsien Wu, Jui-Feng Yeh, E. Hovy
Word sense annotated corpora are useful resources for many text mining applications. Such corpora are only useful if their annotations are consistent. Most large-scale annotation efforts take special measures to reconcile inter-annotator disagreement. To date, however, nobody has investigated how to automatically determine exemplars in which the annotators agree but are wrong. In this paper, we use OntoNotes, a large-scale corpus of semantic annotations, including word senses, predicate-argument structure, ontology linking, and coreference. To determine the mistaken agreements in word sense annotation, we employ word sense disambiguation (WSD) to select a set of suspicious candidates for human evaluation. Experiments are conducted from three aspects (precision, cost-effectiveness ratio, and entropy) to examine the performance of WSD. The experimental results show that WSD is most effective in identifying erroneous annotations for highly-ambiguous words, while a baseline is better for other cases. The two methods can be combined to improve the cleanup process. This procedure allows us to find approximately 2% of the remaining erroneous agreements in the OntoNotes corpus. A similar procedure can be easily defined to check other annotated corpora.
{"title":"Corpus Cleanup of Mistaken Agreement Using Word Sense Disambiguation","authors":"Liang-Chih Yu, Chung-Hsien Wu, Jui-Feng Yeh, E. Hovy","doi":"10.30019/IJCLCLP.200812.0002","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200812.0002","url":null,"abstract":"Word sense annotated corpora are useful resources for many text mining applications. Such corpora are only useful if their annotations are consistent. Most large-scale annotation efforts take special measures to reconcile inter-annotator disagreement. To date, however, nobody has investigated how to automatically determine exemplars in which the annotators agree but are wrong. In this paper, we use OntoNotes, a large-scale corpus of semantic annotations, including word senses, predicate-argument structure, ontology linking, and coreference. To determine the mistaken agreements in word sense annotation, we employ word sense disambiguation (WSD) to select a set of suspicious candidates for human evaluation. Experiments are conducted from three aspects (precision, cost-effectiveness ratio, and entropy) to examine the performance of WSD. The experimental results show that WSD is most effective in identifying erroneous annotations for highly-ambiguous words, while a baseline is better for other cases. The two methods can be combined to improve the cleanup process. This procedure allows us to find approximately 2% of the remaining erroneous agreements in the OntoNotes corpus. A similar procedure can be easily defined to check other annotated corpora.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"27 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115990115","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In recent years, the hierarchical taxonomy integration problem has obtained considerable attention in many research studies. Many types of implicit information embedded in the source taxonomy are explored to improve the integration performance. The semantic information embedded in the source taxonomy, however, has not been discussed in previous research. In this paper, an enhanced integration approach called SFE (Semantic Feature Expansion) is proposed to exploit the semantic information of the category-specific terms. From our experiments on two hierarchical Web taxonomies, the results show that the integration performance can be further improved with the SFE scheme.
{"title":"Hierarchical Taxonomy Integration Using Semantic Feature Expansion on Category-Specific Terms","authors":"Cheng-Zen Yang, Ing-Xiang Chen, Cheng-Tse Hung, Ping-Jung Wu","doi":"10.30019/IJCLCLP.200812.0003","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200812.0003","url":null,"abstract":"In recent years, the hierarchical taxonomy integration problem has obtained considerable attention in many research studies. Many types of implicit information embedded in the source taxonomy are explored to improve the integration performance. The semantic information embedded in the source taxonomy, however, has not been discussed in previous research. In this paper, an enhanced integration approach called SFE (Semantic Feature Expansion) is proposed to exploit the semantic information of the category-specific terms. From our experiments on two hierarchical Web taxonomies, the results show that the integration performance can be further improved with the SFE scheme.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"46 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124011093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-12-01DOI: 10.30019/IJCLCLP.200812.0004
Jen-Liang Chou, Shih-Hung Wu
Wikipedia is the world's largest collaboratively edited source of encyclopedic knowledge. Wikibook is a sub-project of Wikipedia that is intended to create a book that can be edited by various contributors, similar to how Wikipedia is composed and edited. Editing a book, however, requires more effort than editing separate articles. Therefore, methods of quickly prototyping a book is a new research issue. In this paper, we investigate how to automatically extract content from Wikipedia and generate a prototype of a Wikibook as a start point for further editing. Applying search technology, our system can retrieve relevant articles from Wikipedia. A table of contents is built automatically and is based on a two-stage searching method. Our experiments show that, given a keyword as the title of a book, our system can generate a table of contents, which can be treated as a prototype of a Wikibook. Such a system can help free textbook editing. We propose an evaluation method based on the comparison of system results to a traditional textbook and show the coverage of our system.
{"title":"Automatic Wikibook Prototyping via Mining Wikipedia","authors":"Jen-Liang Chou, Shih-Hung Wu","doi":"10.30019/IJCLCLP.200812.0004","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200812.0004","url":null,"abstract":"Wikipedia is the world's largest collaboratively edited source of encyclopedic knowledge. Wikibook is a sub-project of Wikipedia that is intended to create a book that can be edited by various contributors, similar to how Wikipedia is composed and edited. Editing a book, however, requires more effort than editing separate articles. Therefore, methods of quickly prototyping a book is a new research issue. In this paper, we investigate how to automatically extract content from Wikipedia and generate a prototype of a Wikibook as a start point for further editing. Applying search technology, our system can retrieve relevant articles from Wikipedia. A table of contents is built automatically and is based on a two-stage searching method. Our experiments show that, given a keyword as the title of a book, our system can generate a table of contents, which can be treated as a prototype of a Wikibook. Such a system can help free textbook editing. We propose an evaluation method based on the comparison of system results to a traditional textbook and show the coverage of our system.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117026357","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-12-01DOI: 10.30019/IJCLCLP.200812.0001
Yunming Ye, Hongbo Li, Xiaobai Deng, J. Huang
Search interface detection is an essential task for extracting information from the hidden Web. The challenge for this task is that search interface data is represented in high-dimensional and sparse features with many missing values. This paper presents a new multi-classifier ensemble approach to solving this problem. In this approach, we have extended the random forest algorithm with a weighted feature selection method to build the individual classifiers. With this improved random forest algorithm (IRFA), each classifier can be learned from a weighted subset of the feature space so that the ensemble of decision trees can fully exploit the useful features of search interface patterns. We have compared our ensemble approach with other well-known classification algorithms, such as SVM, C4.5, Naive Bayes, and original random forest algorithm (RFA). The experimental results have shown that our method is more effective in detecting search interfaces of the hidden Web.
{"title":"Feature Weighting Random Forest for Detection of Hidden Web Search Interfaces","authors":"Yunming Ye, Hongbo Li, Xiaobai Deng, J. Huang","doi":"10.30019/IJCLCLP.200812.0001","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200812.0001","url":null,"abstract":"Search interface detection is an essential task for extracting information from the hidden Web. The challenge for this task is that search interface data is represented in high-dimensional and sparse features with many missing values. This paper presents a new multi-classifier ensemble approach to solving this problem. In this approach, we have extended the random forest algorithm with a weighted feature selection method to build the individual classifiers. With this improved random forest algorithm (IRFA), each classifier can be learned from a weighted subset of the feature space so that the ensemble of decision trees can fully exploit the useful features of search interface patterns. We have compared our ensemble approach with other well-known classification algorithms, such as SVM, C4.5, Naive Bayes, and original random forest algorithm (RFA). The experimental results have shown that our method is more effective in detecting search interfaces of the hidden Web.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127056356","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-09-01DOI: 10.30019/IJCLCLP.200809.0001
Shu-Ling Huang, Keh-Jiann Chen
In order to train machines to ‘understand’ natural language, we propose a meaning representation mechanism called E-HowNet to encode lexical senses. In this paper, we take interrogatives as examples to demonstrate the mechanisms of semantic representation and composition of interrogative constructions under the framework of E-HowNet. We classify the interrogative words into five classes according to their query types, and represent each type of interrogatives with fine-grained features and operators. The process of semantic composition and the difficulties of representation, such as word sense disambiguation, are addressed. Finally, machine understanding is tested by showing how machines derive the same deep semantic structure for synonymous sentences with different surface structures.
{"title":"Knowledge Representation and Sense Disambiguation for Interrogatives in E-HowNet","authors":"Shu-Ling Huang, Keh-Jiann Chen","doi":"10.30019/IJCLCLP.200809.0001","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200809.0001","url":null,"abstract":"In order to train machines to ‘understand’ natural language, we propose a meaning representation mechanism called E-HowNet to encode lexical senses. In this paper, we take interrogatives as examples to demonstrate the mechanisms of semantic representation and composition of interrogative constructions under the framework of E-HowNet. We classify the interrogative words into five classes according to their query types, and represent each type of interrogatives with fine-grained features and operators. The process of semantic composition and the difficulties of representation, such as word sense disambiguation, are addressed. Finally, machine understanding is tested by showing how machines derive the same deep semantic structure for synonymous sentences with different surface structures.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116135423","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2008-09-01DOI: 10.30019/IJCLCLP.200809.0005
Shih-Hung Liu, Fang-Hui Chu, Yueng-Tien Lo, Berlin Chen
This paper considers minimum phone error (MPE) based discriminative training of acoustic models for Mandarin broadcast news recognition. We present a new phone accuracy function based on the frame-level accuracy of hypothesized phone arcs instead of using the raw phone accuracy function of MPE training. Moreover, a novel data selection approach based on the frame-level normalized entropy of Gaussian posterior probabilities obtained from the word lattice of the training utterance is explored. It has the merit of making the training algorithm focus much more on the training statistics of those frame samples that center nearly around the decision boundary for better discrimination. The underlying characteristics of the presented approaches are extensively investigated, and their performance is verified by comparison with the standard MPE training approach as well as the other related work. Experiments conducted on broadcast news collected in Taiwan demonstrate that the integration of the frame-level phone accuracy calculation and data selection yields slight but consistent improvements over the baseline system.
{"title":"Improved Minimum Phone Error based Discriminative Training of Acoustic Models for Mandarin Large Vocabulary Continuous Speech Recognition","authors":"Shih-Hung Liu, Fang-Hui Chu, Yueng-Tien Lo, Berlin Chen","doi":"10.30019/IJCLCLP.200809.0005","DOIUrl":"https://doi.org/10.30019/IJCLCLP.200809.0005","url":null,"abstract":"This paper considers minimum phone error (MPE) based discriminative training of acoustic models for Mandarin broadcast news recognition. We present a new phone accuracy function based on the frame-level accuracy of hypothesized phone arcs instead of using the raw phone accuracy function of MPE training. Moreover, a novel data selection approach based on the frame-level normalized entropy of Gaussian posterior probabilities obtained from the word lattice of the training utterance is explored. It has the merit of making the training algorithm focus much more on the training statistics of those frame samples that center nearly around the decision boundary for better discrimination. The underlying characteristics of the presented approaches are extensively investigated, and their performance is verified by comparison with the standard MPE training approach as well as the other related work. Experiments conducted on broadcast news collected in Taiwan demonstrate that the integration of the frame-level phone accuracy calculation and data selection yields slight but consistent improvements over the baseline system.","PeriodicalId":436300,"journal":{"name":"Int. J. Comput. Linguistics Chin. Lang. Process.","volume":"305 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2008-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123120802","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：