Dagan, Ido, Alon Itai, and Ulrike Schwall (1991). "Two Languages are More Informative than One", in Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (ACL91). Berkeley.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....have sufficient statistics that are lower order marginal distributions. In the future, we will investigate other goodness of fit tests ( 18] 1] 22] that are perhaps more appropriate for sparse data. The Experiment Unlike several previous approaches to word sense disambiguation ( 29] 5] [7], 10] nothing in this approach limits the selection of sense tags to a particular number or type of meaning distinctions. In this study, our goal was to address a non trivial case of ambiguity, but one that would allow some comparison of results with previous work. As a result of these ....

....Comparison to Previous Work Many researchers have avoided characterizing the interactions among multiple contextual features by considering only one feature in determining the sense of an biguous word. Techniques for identifying the optimum feature to use in disambiguating a word are presented in [7], 30] and [S] Other works consider multiple con textual features in performing disambiguation without formally characterizing the relationships among the features. The majority of these efforts ( lSl, all) weight each feature in predicting the sense of an ambiguous word in accordance with ....

Dagan, I., Itai, A., and Schwall, U. (1991). Two Languages Are More Informative Than One. Proceedings of the 2gth Annual Meeting of the Asso- ciation for Computational Linguistics (ACL-91), pp. 130-137.

....the target language documents. As was found with monolingual concept hierarchies, their quality and richness can be improved by including terms found within the documents in addition to those of the query. The accurate translation of the additional terms will be conducted using Dagans technique [5], which is designed to work with minimal translation resources. When translating a particular term, the context in which it occurs is used to disambiguate the term. If that term occurs in other retrieved documents, it is reasonable to assume it will be used in the same sense throughout those ....

Dagan, I., Itai, A. and Schwall, U. Two languages are more informative than one. Proceedings of the ACL(29): 130-137, 1991

....sufficient statistics that are lower order marginal distributions. In the furture, we will investigate other goodness of fit tests ( 18] 1] 22] that are perhaps more appropriate for sparse data. The Experiment Unlike several previous approaches to word sense disam biguation ( 29] 5] [7], 10] nothing in this approach limits the selection of sense tags to a particular number or type of meaning distinctions. In this study, our goal was to address a non trivial case of ambiguity, but one that would allow some comparison of results with previous work. As a result of these ....

....Comparison to Previous Work Many researches have avoided characterizing the interactions among multiple contextual features by considering only one feature in determining the sense of an biguous word. Techniques for identifving the optimum feature to use in disambiguating a word are presented in [7], 30] and [5] Other works consider multiple con textual features in performing disambiguation without formally characterizing the relationships among the fea tures. The majority of these efforts ( 13] 31] weight each feature in predicting the sense of an ambiguous word in accordance with ....

Dagan, I., Itai, A., and Schwall, U. (1991). Two Languages Are More Informative Than One. Proceedings of the 29th Annual Meeting of the Asso- ciation for Computational Linguistics (A CL-91), pp. 130-137.

....same collection is being used for both disambiguation and retrieval, so domain relevance of the filtering process is guaranteed. In this light, the translation disambiguation problem bears a strong resemblence to term disambiguation in a 18 monolingual setting. In fact, a number of researchers [8, 3] have used cross language relationships to help with disambiguation. Given the limited success of term disambiguation as a tool for IR [23] there is some question about whether we can hope to gain any benefits out of translation filtering. Translation disambiguation may well work better as an ....

I. Dagan, A.Itai, and U. Schwall. Two languages are more informative than one. In Proc. of the Association for Computational Linguistics (ACL), pages 130--137, 1991. 19

....a probabilistic classifier for word sense disambiguation did not attempt to systematically identify the interdependencies among contextual features that can be used to classify the meaning of an ambiguous word. Many researchers have performed disambiguation on the basis of only a single feature ([6], 15] 2] while others who do consider multiple contextual features assume that all contextual features are either conditionally independent given the sense of the word ( 8] 14] or fully independent ( 10] 16] In earlier work, we describe a method for identifying an appropriate model ....

....the presence or absence of a word that is one of the four most frequently occurring content words in a data sample composed of sentences containing the word to be disambiguated. This strategy for selecting collocation specific variables is simpler than that used by many other researchers ([6], 15] 2] This simpler method was chosen to support work we plan to do in the future (eliminating the need for sense tagged data; see section 6) In using this strategy, we do, however, run the risk of reducing the informativeness of the variables. With the variables as described above, the ....

Dagan, I.; Itai, A.; and Schwall, U. (1991). Two Languages Are More Informative Than One. Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (ACL-91), pp. 130-137.

.... 1993, Voorhees, 1993, Richardson, 1994) These three authors all used the WordNet lexical database (Beckwith, Fellbaum, Gross and Miller, 1992) In recent years there has been much interest in statistical word sense disambiguation using corpora (Brown, Della Pietra, Della Pietra and Mercer, 1991; Dagan, Itai and Schwall, 1991; Zernik, 1991; Yarowsky, 1992; Gale, Church and Yarowsky, 1993; Schutze, 1993) In general, these methods gather statistical co occurrence data for each word by examining a large number of contexts obtained from one or more text corpora. As a result it is possible to cluster the usages by word ....

Dagan, I., Itai, A., & Schwall, U. (1991). Two Languages are more Informative than One. Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, Berkeley, California, 18-21 June, 1991, 130-137.

....have sufficient statistics that are lower order marginal distributions. In the furture, we will investigate other goodness of fit tests ( 18] 1] 22] that are perhaps more appropriate for sparse data. The Experiment Unlike several previous approaches to word sense disambiguation ( 29] 5] [7], 10] nothing in this approach limits the selection of sense tags to a particular number or type of meaning distinctions. In this study, our goal was to address a non trivial case of ambiguity, but one that would allow some comparison of results with previous work. As a result of these ....

....to Previous Work Many researches have avoided characterizing the interactions among multiple contextual features by considering only one feature in determining the sense of an ambiguous word. Techniques for identifying the optimum feature to use in disambiguating a word are presented in [7], 30] and [5] Other works consider multiple contextual features in performing disambiguation without formally characterizing the relationships among the features. The majority of these efforts ( 13] 31] weight each feature in predicting the sense of an ambiguous word in accordance with ....

Dagan, I., Itai, A., and Schwall, U. (1991). Two Languages Are More Informative Than One. Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (ACL-91), pp. 130-137.

....These techniques rely on parallel corpora and our interest is in ascertaining whether unlinked corpora can be used effectively for phrasal translation. Kraaij and Hiemstra [KH97] used co occurrence frequency for phrase translation with some success during the TREC 6 [Har97] evaluations. In [DIS91] a co occurrence method was used for target word selection, however there have been no reports of its use in a retrieval environment. A description of our method follows. Given two tagged source terms, collect all target translation equivalents appropriate to each term s part of speech. Generate ....

Ido Dagan, Alon Itai, and Ulrike Schwall. Two languages are more informative than one. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pages 130--137, 1991.

....Relations to Other Work Many researches have avoided characterizing the interactions among multiple contextual features by considering only one feature in determining the sense of an ambiguous word. Techniques for identifying the optimum feature to use in disambiguating a word are presented in [22], 73] and [10] Other works consider multiple contextual features in performing disambiguation without formally characterizing the relationships among the features. The majority of these efforts [34] 76] weight each feature in predicting the sense of an ambiguous word in accordance with ....

Dagan, I., Itai, A., and Schwall, U. (1991). Two Languages Are More Informative Than One. Procs of the 29th Annual Meeting of the Association for Computational Linguistics (ACL-91), pp. 130-137.

.... literature also reflects an increasing interest in statistical training methods for many NL tasks, including parsing [Jelinek and Lafferty 1991, Magerman and Marcus 1991, Bobrow 1991, Magerman and Weir 1992, Black, Jelinek, et al. 1992] part of speech tagging [Church 1988] and corpora alignment [Dagan et al. 1991, Gale and Church 1991] Simply stated, we seek to build a parser that can construct accurate syntactic and semantic analyses for the sentences of a given language. The parser should know little or nothing about the target language, save what it can discover statistically from a representative ....

Dagan, I., Itai, A., Schwall, U. "Two Languages Are More Informative Than One," Proc. of the 29th Annual Meeting of the Association for Computational Linguistics. Berkeley, California, 1991, pp. 130-137.

....among feature variables, Bruce and Wiebe, 1994b) introduce the use of sequential model selection and decomposable models for word sense disambiguation. 5 Alternative probabilistic approaches have involved using a single contextual feature to perform disambiguation (e.g. Brown et al. 1991) (Dagan et al. 1991), and (Yarowsky, 1993) present techniques for identifying the optimal feature to use in disambiguation) Maximum Entropy models have been used to express the interactions among multiple feature variables (e.g. Berger et al. 1996) but within this framework no systematic study of interactions ....

I. Dagan, A. Itai, and U. Schwall. 1991. Two languages are more informative than one. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pages 130--137.

....monolingual) knowledge. In the field of computational linguistics, a variety of researches have investigated the use of bilingual corpora, including sentence alignment [Wu 1994] word correspondence [Dagan 1993] collocation correspondence [Smadja 1996] word sense disambiguation [Brown 1991b, Dagan 1991, Gale 1992] and machine translation [Brown 1990, Su 1995, Wu 1995] Bi lingual material is also valuable for linguists who are interested in comparative studies of different languages, or in doing bi lingual lexicography for translation. Here, we introduce three kinds of techniques which can ....

Dagan, I., A. Itai, U. Schwall, "Two Language Are More Informative Than One," Proceedings of ACL-29, pp. 130-137, California, USA, June 18-21, 1991.

....by letting this distinction be determined by an external knowledge source such as a thesaurus or a dictionary. Although this requirement may seem trivial, most corpus based methods do not, in fact, allow such flexibility. For example, defining the senses by the possible translations of the word (Dagan and Itai, 1991; Brown et al. 1991; Gale, Church, and Yarowsky, 1992) by the Roget s categories (Yarowsky, 1992) or by clustering (Schutze, 1992) yields a grouping that does not always conform to the desired sense distinctions. In comparison, our reliance on the MRD the definition of senses in the ....

....the 5 Resnik (1995) recently suggested to overcome this particular difficulty by a different measure that takes into account the informativeness of the most specific common ancestor of the two words. need for manual work, as producing bilingual corpora requires manual translation work. 6 Dagan and Itai (1991) used a bilingual lexicon and a monolingual corpus, to save the need for translating the corpus. The problem remains, however, that the word translations do not necessarily overlap with the desired sense distinctions. Schutze (1992) clustered the examples in the training set, and manually assigned ....

Dagan, I. and A. Itai. 1991. Two languages are more informative than one. In Proceedings of the 29th Annual Meeting of the ACL, pages 130--137.

.... online corpus and ranking possible word senses according to the degree of match between sense evidence and the context of the ambiguous word: Cowie [4] Wallis [24] and Demetriou [8] have made further use of dictionaries; Zernik [27] built a disambiguator using a morphological analyser; Dagan [7] used bilingual corpora; Church [3] tried aligned bilingual corpora; Voorhees [23] and Sussna [20] used the WordNet thesaurus; and Yarowsky [26] used a combination of Roget s thesaurus and Grollier s encyclopaedia to produce one of the better performing disambiguators to date, achieving 90 ....

I. Dagan, A. Itai, and U. Schwall. Two languages are more informative than one. In Proceedings of the ACL, pages 130--137, 1991.

....sophisticated approaches to MT. However, they are only useful if they can be discovered reliably and efficiently. Their time may come when we have a better understanding of how to model the human translation process. 10 Conclusion It is well known that two languages are more informative than one (Dagan et al. 1991). I have argued that texts in two languages are not only preferable but necessary for discovery of noncompositional compounds for translation related applications. Given a method for constructing statistical translation models, NCCs can be discovered by maximizing the models information theoretic ....

I. Dagan, A. Itai & U. Schwall. (1991) "Two Languages are More Informative than One," Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics. Berkeley, CA.

....in detail here. For pointers see [Ng and Zelle1997] Chandrasekar and Bangalore1998] Kilgarriff1997] Resnik1997] A variety of methods has been used, such as supervised learning from large corpora, e.g. Yarowsky1995] multilingual approaches using the difference between languages. e.g. [Dagan et al..1991], statistical models, e.g. Bruce and Wieber1994] and using linguistic and contextual information, e.g. Ng and Lee1996] Previous approaches to the extraction of semantic information have tended to use general language dictionaries or thesauri. One common method is to compare the content of the ....

I. Dagan, A. Itai, and U. Schwall. 1991. Two languages are more informative than one. In Proc. of the Association for Computational Linguistics (ACL), pages 130--137.

....be improved. Other candidate improvements were suggested in earlier sections. 7.1 Bootstrapping from Bilingual Corpora The time required for hand labeling the training sentences is prohibitive, but there is a way it might be automated. Recently several researchers (e.g. Brown et al. 1991) (Dagan et al. 1991), Inoue 1991) have suggested using bilingual aligned corpora in the lexical disambiguation task, the term aligned indicates that within the bilingual database, sentences that are translations of one another are grouped together) Although most of the discussion is in terms of choosing words ....

Dagan, I., A. Itai, & U. Schwall (1991). Two languages are more informative than one. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pages 130--137.

....identified clause alignments. Introduction The availability of large collections of texts in electronic form, has given rise to a wide range of applications aiming at the elicitation of linguistic resources such as translation dictionaries, transfer grammars and retrieval of translation examples (Dagan et al. 1991; Matsumoto et al. 1993) or even the building of fully blown machine translation systems (Brown et al. 1990) The purpose of this paper is to describe a technique for extracting translation correspondences at bellow sentence level by employing statistical techniques coupled with shallow ....

Dagan, I., A. Itai, and U. Schwall.1991. Two languages are more informative than one. In Proc. 29 th Annual Meeting of the Association for Computational Linguistics, 130-137.18-21 June, Berkley, Calif.

....P (A i jw) 1. For reasons which will be elaborated in Section 2, our problem is most acute in Hebrew and some other languages (e.g. Arabic) Though ambiguity problems of a similar nature occur in other languages. One such problem is sense disambiguation. In the context of Machine Translation, Dagan and Itai (Dagan, Itai, and Schwall, 1991; Dagan and Itai, 1994) used corpora in the target language to resolve ambiguities in the source language. Yarowsky (Yarowsky, 1992) proposed a method for sense disambiguation using wide contexts. Part of speech tagging deciding the correct part of speech in the current context of the sentence ....

....P i Gamma1 j j ) Figure 1 Calculating the approximated morpho lexical probabilities. 9 Another potentially problematic case is the coverage problem, which arises whenever we do not have enough data in the corpus for disambiguation of a certain word (see a discussion on this problem in (Dagan, Itai, and Schwall, 1991)) This problem was found to occur very rarely for only 3 of the ambiguous words in our test texts the counters found in the corpus were smaller than 20. We expect this percentage would be even smaller had we used a larger training corpus. For such words, we simply ignored the data and ....

Dagan, Ido, Alon Itai, and Ulrike Schwall. 1991. Two languages are more informative than one. In Proc. of the Annual Meeting of the ACL, pages 130--137.

No context found.

Ido Dagan, Alon Itai, and Ulrike Schwall. 1991. Two languages are more informative than one. In Proc. of the Annual Meeting of the ACL, pages 130--137.

No context found.

Dagan, Ido, Alon Itai, and Ulrike Schwall (1991). "Two Languages are More Informative than One", in Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (ACL91). Berkeley.

No context found.

I. Dagan, A. Itai, and U. Schwall. Two languages are more informative than one. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pp. 130-137, 1991.

No context found.

Ido Dagan, Alon Itai, and Ulrike Schwall. Two languages are more informative than one. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pages 130--137, 1991.

No context found.

Dagan, I., A. Ital and U. Schwall, "Two Languages are more Informative than One," Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pp. 130-137, 1991.

No context found.

I. Dagan, A. Itai, and U. Schwall. Two languages are more informative than one. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pp. 130-137, 1991.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC