E. Sumita and H. Iida. 1991. Experiments and Prospects of Example-Based Machine Translation. In Proceedings of the 29th Annual Meeting of the Association of Computational Linguistics (ACL).  Home/Search   Document Details and Download   Summary   Related Articles   Check

....which is composed of two subsystems: learning of translation templates from examples and translation based on template matching. This paper discusses in particular how to learn translation templates from examples. While most previous research in this area has focused on other aspocts, Sato90][Sumita91] we believe that automatic learning from examples is essential for implementing practical example basod machine translation systems. ilingualh orpus. e Lang TTamn; TRANSLATION ] Fig.l Two Phase Example Based Machine Translation ALrrES BE COLING 92, NANTES, 23 28 Aor 1992 6 7 2 PROC. OF ....

Stunira, E. and H. Iida: "Experiments and Prospects of Example-based Machine Translation", Proc. of 29th Annual Meeting of the ACL, pp,185~192 (June 1991). Acres DE COLING-92, N^NTES, 23-28 ^or 1992 6 7 8 Paoc. OF COLING-92, NANTES, AUG. 23-28, 1992

....input This research has been supported in part by NATO Science for Stability Program Grant TU LANGUAGE. sentence, the correspondences between the source and target languages should be available to the system; however this issue has not been given enough consideration by the current EBMT systems [3, 5, 10, 11, 14]. Kitano has adopted the manual encoding of the translation rules, however this is a difficult and an error prone task for a large corpus. In this paper, we formulate this acquisition problem as a machine learning task in order to automate the process. We use example based learning techniques to ....

Sumita, E. and Iida, H.: Experiments and Prospects of Example-Based Machine Translation, Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (1991). This article was processed using the L a T E X macro package with LLNCS style

....user interface for grammar debugging. 1 Introduction As the area of the task becomes wider and the task itself gets more complex, natural language interface becomes more indispensable in order to brief the output of the system. On the other hands, example based natural language processing [SN90, SI91] is proposed by recent researches, which is one of the promising approaches to offer better natural language interface. Example based natural language processing, however, requires large corpora especially specific to each task domain. Constructing such corpora is laborious and needs various sorts ....

Eiichiro SUMITA and Hitoshi IIDA. "Experiments and Prospects of Example-based Machine Translation". In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (ACL), pages 185--192, 1991.

....the last few years this situation has changed considerably. In particular in Japan there have been extensive efforts regarding the automatic translation from Japanese into English and vice versa. One of the most prominent research directions in Japan has been example based machine translation [9, 10, 11], which relies on massive bilingual corpora to build a knowledge base of translation examples. New sentences are then translated by finding the most similar example. Unfortunately, this promising approach can only be applied successfully to language pairs for which enough bilingual data is ....

E. Sumita and H. Iida, "Experiments and Prospects of Example-Based Machine Translation", Proc. of the Annual Meeting of the Association for Computational Linguistics, 1991, pp. 185-192.

....for the sentence in the source language, parts of the corresponding target language sentence are constructed using structural equivalences and deviances in those matches. Following Nagoa s original proposal, several machine translation methods that utilize bilingual corpora have been studied [5, 9, 20, 21, 22, 23]. Some researchers [3, 24] only utilize bilingual corpora to create a bilingual dictionary and use it during the translation process. In other words, they aligned bilingual corpora at word level to figure out corresponding words in languages. Bilingual corpora is also aligned at phrase level by ....

Sumita, E., and Iida, H., Experiments and Prospects of Example-Based Machine Translation, in: Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, 1991.

....to other language pairs by making extensive reuse of the already available linguistic knowledge. At the sentence level we follow the transfer based translation paradigm by applying the translation rules learnt from the examples provided by the user. Unlike pure example based machine translation [13, 18, 21] we do not compare new sentences with all previous translation examples but we abstract from the examples by generating translation rules. However, we still keep a record of all relevant examples to verify the effect of a new translation rule on other examples. This is necessary to guarantee the ....

SUMITA, E. and IIDA, H., Experiments and Prospects of Example-Based Machine Translation, in: Proc. of the Annual Meeting of the Association for Computational Linguistics 1991.

....for the sentence in the source language, parts of the corresponding target language sentence are constructed using structural equivalences and deviances in those matches. Following Nagoa s original proposal, several machine translation methods that utilize bilingual corpora have been studied [5, 9, 16, 17, 18, 19]. Some researchers [3, 20] only utilized bilingual corpora to create a bilingual dictionary and use it during the translation process. In other words, they aligned bilingual corpora at word level to figure out corresponding words in languages. Bilingual corpora is also aligned at phrase level by ....

Sumita, E., and Iida, H., Experiments and Prospects of Example-Based Machine Translation, in: Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, 1991.

....for the sentence in the source language, parts of the corresponding target language sentence are constructed using structural equivalences and deviances in those matches. Following Nagoa s original proposal, several machine translation methods that utilize bilingual corpora have been studied [5, 12, 22, 23, 24, 25]. Some researchers [3, 26] only utilized bilingual corpora to create a bilingual dictionary and use it during the translation process. In other words, c fl 2000 Kluwer Academic Publishers. Printed in the Netherlands. APIN66499.tex; 3 07 2000; 10:03; p.1 2 Cicekli and Guvenir they aligned ....

Sumita, E., and Iida, H., Experiments and Prospects of Example-Based Machine Translation, in: Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, 1991.

....I. CICEKLI EBMT has been proposed by Nagao [14] as translation by analogy which is in parallel with memory based reasoning [20] case based reasoning [16] and derivational analogy [2] Examplebased translation relies on the use of past translation examples to derive a translation for a given input [3, 9, 19, 21]. The input sentence to be translated is compared with the example translations analogically to retrieve the closest examples. Then, the fragments of the retrieved examples are translated and recombined in the target language. Prior to the translation of an input sentence, the correspondences ....

Eiichiro Sumita and Hitoshi Iida. Experiments and prospects of example-based machine translation. In 29th Annual Meeting of the Association for Computational Linguistics. Proceedings of the Conference. 18-21 June 1991. University of California, Berkeley, California, USA, pp. 185-192 (1991).

....90] and [Kaji 92] have tackled the problem of establishing correspondences between words and phrases in bilingual texts. The third key issue of EBMT, that is exploiting the retrieved translation example, is usually dealt with by integrating into the system conventional MT techniques [Kaji 92] Sumita 91] Simple modifications of the translation proposal, such as word substitution, would also be possible, provided that alignment of the translation archive at word level was available. In establishing a mechanism for the best match retrieval, which is the topic of this paper, the crucial tasks ....

....words of the two sentences in terms of their morphological paradigms, synonyms, hyperonyms, hyponyms, antonyms, pos tags. Nirenburg 93] or use a semantic distance d (0 d 1) which is determined by the Most Specific Common Abstraction (MSCA) obtained from a thesaurus abstraction hierarchy [Sumita 91] Then, a similarity metric is devised, which reflects the similarity of two sentences, by combining the individual contributions towards similarity stemming from word comparisons. The word based metrics are the most popular, but other approaches include syntax rule driven metrics [Sumita 88] ....

Sumita E. and H. Iida, (1991). "Experiments and Prospects of Example-based Machine Translation". Proc. of the 29th Annual Meeting of the Association for Computational Linguistics, pp 185-192.

....they really require a more specialised source of information. To this extent, we use information from a domain specific thesaurus to uncover semantic relations concerning candidate terms in the text. We propose a new similarity measure, based on methods more commonly found in machine translation [14] [17] to calculate semantic distance between terms and their context, as will be detailed in Section . 3 Extracting Contextual Information It is well known among terminologists that linguistic contexts provide a rich and valuable source of information. The problem lies in identifying those parts ....

....of d as a context term of a. 3. 2 Similarity Weight Unlike other context similarity weights, e.g. 16] 12] our weight is based not on statistical but on semantic properties of the context terms [9] Similarity is calculated using a thesaurus hierarchy, similar to the approach of Sumita and Iida [14]. For this we use the UMLS Metathesaurus to assign semantic categories to each word in question. The similarity between a context term and a candidate term is then measured by calculating the distance between their semantic categories in the hierarchy provided by the UMLS Semantic Network, ....

E. Sumita and H. Iida. Experiments and prospects of example-based machine translation. In Proc. of 29th Annual Meeting of the Association for Computational Linguistics, pages 185--192, Berkeley, California, 1991.

....future work. 4.2 Calculating Similarity in the UMLS Semantic Network The similarity between context term and term is measured using the UMLS Semantic Network, which is a hierarchy of semantic types. Our approach is modelled on techniques used in Example Based Machine Translation (EBMT) 21] [19]. EBMT aims to select the most similar example for a given problem, by defining a set of relevant examples and comparing them with the problem in terms of context and configuration. Most of these comparison methods involve using some kind of semantic distance measure, based on the relative ....

....paths of the two words back up the hierarchy until a common ancestor is found, and then measuring the average distance from node to MSCA. The shorter the distance to the MSCA, the more similar the two words. The position of the MSCA within the hierarchy can also be used to measure similarity [21] [19]. The lower down in the hierarchy the MSCA, the more specific it is and therefore ORGANISM PHYSICAL OBJECT ANATOMICAL STRUCTURE STRUCTURE EMBRYONIC CONCEPTUAL ENTITY [TA] ENTITY [TA1] TA2] TA11] TA12] TA112] FUNGUS [TA111] PLANT [TA121] TA122] ANATOMICAL ABNORMALITY [TA1111] ALGA Figure 1: ....

E. Sumita and H. Iida. Experiments and prospects of example-based machine translation. In Proc. of 29th Annual Meeting of the Association for Computational Linguistics, pages 185--192, Berkeley, California, 1991.

.... This research has been supported in part by NATO Science for Stability Program Grant TU LANGUAGE. sentence, the correspondences between the source and target languages should be available to the system; however this issue has not been given enough consideration by the current EBMT systems [3, 5, 10, 11, 14]. Kitano has adopted the manual encoding of the translation rules, however this is a difficult and an error prone task for a large corpus. In this paper, we formulate this acquisition problem as a machine learning task in order to automate the process. We use example based learning techniques to ....

Sumita, E. and Iida, H.: Experiments and Prospects of Example-Based Machine Translation, Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (1991). This article was processed using the L a T E X macro package with LLNCS style

.... the calibration test on a large number of ratio combinations, to choose the one which leads to the optimum correspondence with the results obtained using the control metric; d) augmenting the set of comparison criteria, for instance, by including membership in the same semantic class (as did Sumita and Iida, 1991), though this criterion is inherently weaker than synonymy or direct hyperonymy. This enhancement presupposes the availability of a thesaurus or another source of semantic class markers. Yet another major area of improvement has to do with creating a complete experimental set up which will allow ....

Sumita, E. and H. Iida. 1991. Experiments and prospects of example-based machine translation.

....sense and general knowledge. The only major new lines of investigation that have emerged in recent years have involved the use of existing translations as a prime source of information for the production of new ones. One form that this takes is that of example based machine translation (Furuse Iida, 1992; Iida Iida, 1991; Nagao, 1992; Sato, 1992) in which a system of otherwise fairly conventional design is able to refer to a collection of existing translations. A much more radical approach, championed by IBM (Brown, Cocke, et al. 1990) is the one in which virtually the entire body of knowledge that the system ....

Iida, E. S. and Iida, H. (1991). Experiments and prospects of example-based machine translation. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pages 185--192, Berkeley, California. Association for Computational Linguistics.

....successful efforts at extracting selectional patterns from corpora [Sekine et al. 92, Chang et al. 92, Grishman and Sterling 92] In the realm of machine translation, there have been two avenues of development. On the one hand, the work on Example Based Machine Translation [Sato and Nagao 90, Sumita and Iida 91] and on Bilingual Knowledge Bases [Sadler and Vendelman 90] has shown that collections of manually selected, syntactically analyzed bilingual examples can be an effective source of translation information, substituting for explicitly prepared transfer rules. On the other hand, the 3 work on ....

E. Sumita and H. Iida. Experiments and prospects of examplebased machine translation. Proc. 29th Annl. Meeting Assn. for Computational Linguistics, Berkeley, Ca, 1991. 12

....the load of morphological syntactic generation module with pre defined natural translations for one or more target languages. They also reduce the search space for parser and resolve some POS (part ofspeech) ambiguities [Lee, 1994a] This approach is similar example based MT [Sumita et al. 1990] [Sumita, 1991] in that both try to find translation patterns, but example based approach is for translation itself. On the other hand, compound unit recognizer is for obtaining performance and speed by plug in between a morphological analyzer and a parser. We define compound unit (CU) as a combined concept of ....

E. Sumita. Experiments and Prospects of Example-Based Machine Translation. In Proceedings of 29th Annual Meeting of the Association for Computational Linguistics, 1991

....syntactic coverage. Overall, these problems are caused by a lack of linguistic knowledge installed in parsers. Recently, for the purpose of using information included in raw or tagged text, example based methods utilizing probabilities for grammar rules (Fujisaki et al. 1989) semantic similarity (Sumita and Iida, 1991), and so on have been proposed. Our method, which is one of the example based approaches, analyzes an input sentence by analogy; that is, it captures by certain relationships between the input and similar sentences in a treebank. This technique gives us a new view of examples for disambiguation, ....

E. Sumita and H. Iida. 1991. Experiments and Prospects of Example-Based Machine Translation.

....them with the problem in terms of context and configuration. Most of these comparison methods involve using some kind of semantic distance measure, based on the relative positions of the two items within a hierarchical network. Our approach is modelled on techniques used in EBMT [Zhao1996] [Sumita and Iida1991]. By defining a hierarchically based representation, we can use a similarity measurement based on semantic distance. Like Grefenstette, we measure similarity in terms of shared attributes, although we extend this idea by comparing the similarity of the attributes rather than just calculating the ....

E. Sumita and H. Iida. 1991. Experiments and prospects of example-based machine translation. In Proc. of 29th Annual Meeting of the Association for Computational Linguistics, pages 185--192, Berkeley, California.

....based reasoning [14] case based reasoning [11] and derivational analogy [2] Example based translation relies on the use of past translation examples 1 This research has been supported in part by NATO Science for Stability Program Grant TULANGUAGE. to derive a translation for a given input [3, 9, 12, 13, 15]. The input sentence to be translated is compared with the example translations analogically to retrieve the closest examples to the input. Then, the fragments of the retrieved examples are translated and recombined in the target language. Prior to the translation of an input sentence, the ....

Sumita, E. & Iida, H.: Experiments and Prospects of Example-Based Machine Translation, Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (1991).

....which were semantically similar to the components of the input sentence, then construct the translated sentence by putting together the translations of the input components according to the grammatical rules of the target language. A simple example of this technique was developed by Sumita et al. [75, 74] and called EBMT (for Example Based Machine Translation) This system translated only phrases of the form noun 1 NO noun 2 , where NO is the general partitive adposition in Japanese. In most contexts, it is similar to the English word of, but can also be translated as in (as in the ....

E. Sumita and H. Iida. Experiments and prospects of example-based machine translation. Proceedings of the ACM, 1991.

.... [4, 6, 10] EBMT has been proposed by Nagao [8] as Translation by Analogy which is in parallel with memory based reasoning [14] case based reasoning [11] and derivational analogy [2] Examplebased translation relies on the use of past translation examples to derive a translation for a given input [3, 9, 12, 13, 15]. The input sentence to be translated is compared with the example translations analogically to retrieve the closest examples to the input. Then, the fragments of the retrieved examples are translated and recombined in the target language. Prior to the translation of an input sentence, the ....

Sumita, E. & Iida, H.: Experiments and Prospects of Example-Based Machine Translation, Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (1991).

.... 7, 11] EBMT has been proposed by Nagao [9] as translation by analogy which is in parallel with memory based reasoning [14] case based reasoning [12] and derivational analogy [2] Example based translation relies on the use of past translation examples to derive a translation for a given input [3, 10, 13, 15]. The input sentence to be translated is compared with the example translations analogically to retrieve the closest examples. Then, the fragments of the retrieved examples are translated and recombined in the target language. Prior to the translation of an input 1 This research has been ....

Sumita, E. & Iida, H.: Experiments and Prospects of Example-Based Machine Translation, Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (1991).

.... as translation by analogy which is in parallel with memory based reasoning [20] case based reasoning [16] and derivational analogy [2] Examplebased translation relies on the use of past translation examples to derive a translation for a given 354 Halil Altay G uvenir and Ilyas Cicekli input [3, 9, 19, 21]. The input sentence to be translated is compared with the example translations analogically to retrieve the closest examples. Then, the fragments of the retrieved examples are translated and recombined in the target language. Prior to the translation of an input sentence, the correspondences ....

E. Sumita and H. Iida. Experiments and prospects of example-based machine translation. In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, pp. 185--192, Berkeley, California, USA (1991).

.... For effective and robust spoken language translation, a speech translation system called Transfer Driven Machine Translation (TDMT) which car ries out analysis and translation in an example based framework has been proposed[6] TDMT which refers to as Example Based Machine translation(EBMT)[7] does not require a full analysis and instead defines patterns on sentences phrases expressed by variables and constituent boundaries . These patterns are classified into several classes, for example a complex sentence pattern class, an embedded clause pattern class, and phrase class. A ....

E.Sumita, H.Iida:"Experiments and Prospects of Example-based Machine Translation" In Proc. of 9th ACL, pp.185-192, 1991.

....indicates that the source pattern X K Y corresponds to many possible target patterns. TDMT selects the semantically most similar target pattern, and then translates the input by using the pattern. This is enabled by measuring semantic distance (similarity) in terms of a thesaurus hierarchy [Sumita and Iida, 1991]. The transfer process involves the derivation of possible source structures by a Constituent Boundary parser (CB parser) Furuse and Iida, 1996] and a mapping to target structures. When a structural ambiguity occurs, the best structure is determined according to the total semantic distances of ....

Eiichiro Sumita and Hitoshi Iida. 1991. Experiments and Prospects of Example-based Machine Translation. In Proceedings of the 29th ACL, pages 185--192.

No context found.

E. Sumita and H. Iida. 1991. Experiments and Prospects of Example-Based Machine Translation. In Proceedings of the 29th Annual Meeting of the Association of Computational Linguistics (ACL).

No context found.

Sumita E. and H. Iida, (1991). "Experiments and Prospects of Example-based Machine Translation". Proc. of the 29th Annual Meeting of the Associatiott for Computational Linguistics, pp 185-192.

No context found.

Sumita, E. and H. Iida (1991). "Experiments and Prospects of Example-Based Machine Translation," Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics, University of California, Berkeley, June 18-21.

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