All issues
Volume 1 (1994)
- Issue 1 Pages 1-
Volume 1, Issue 1
Displaying 1-5 of 5 articles from this issue
- |<
- <
- 1
- >
- >|
-
[in Japanese]1994 Volume 1 Issue 1 Pages 1
Published: October 10, 1994
Released on J-STAGE: March 01, 2011
JOURNAL FREE ACCESSDownload PDF (96K) -
SADAO KUROHASHI, MAKOTO NAGAO1994 Volume 1 Issue 1 Pages 3-20
Published: October 10, 1994
Released on J-STAGE: March 01, 2011
JOURNAL FREE ACCESSTo understand a text or dialogue, one must track the discourse structure. Whilework on discourse structure has mainly focused on knowledge employed in the analysis, detailed knowledge with broad coverage availability to computers is unlikely tobe constructed for the present. In this paper, we propose an automatic method fordetecting discourse structure by a variety of keys existing in the surface informationof sentences. We have considered three types of clue information: clue expressions, occurrence of identical/synonymous words/phrases, and similarity between two sentences. Experimental results have shown that, in the case of scientific and technicaltexts, considerable part of the discourse structure can be estimated by incorporatingthe three types of clue information, without performing sentence understandingprocesses by giving knowledge to computers.View full abstractDownload PDF (1602K) -
Kenji Kita, Yasuhiko Kato, Takashi Omoto, Yoneo Yano1994 Volume 1 Issue 1 Pages 21-33
Published: October 10, 1994
Released on J-STAGE: March 01, 2011
JOURNAL FREE ACCESSWhile corpus-based studies are now becoming a new methodology in natural languageprocessing, second language learning offers one interesting potential application. In this paper, we are primarily concerned with the acquisition of collocational knowledge from corpora for use in language learning. First we discuss the importance of collocational knowledge in second language learning, and then take up two measures, mutual information and cost criteria, for automatically identifying or extractingcollocations from corpora. Comparative experiments are made between the two measures using both Japanese and English corpora. In our experiments, the cost criteria measure proved more effective in extracting interesting collocations such as fundamental idiomatic expressions and phrases.View full abstractDownload PDF (1229K) -
SADAO KUROHASHI, MAKOTO NAGAO1994 Volume 1 Issue 1 Pages 35-57
Published: October 10, 1994
Released on J-STAGE: March 01, 2011
JOURNAL FREE ACCESSConventional parsing methods can not analyze long sentences precisely, since thesemethods consider no information in a wide-range series of words in a sentence. Wehave succeeded in developing a method of detecting coordinate structures by a similaritymeasure of two arbitrary series of words. This paper describes a method ofparsing a sentence by using the information of coordinate structures. A long sentencecan be reduced into a shorter form by recognizing coordinate structures init. Consequently the total dependency structure of a sentence can be obtained byrelatively simple modifier/modifiee rules. We report the results of analyzing 150Japanese sentences to illustrate the effectiveness of this method.View full abstractDownload PDF (7188K) -
Hideo Watanabe1994 Volume 1 Issue 1 Pages 59-75
Published: October 10, 1994
Released on J-STAGE: March 01, 2011
JOURNAL FREE ACCESSWhen the example-based approach is used for machine translation, it is importantto collect a large volume of translation patterns, because most systems use as atranslation example a pair of parsed structures in the source and target languages.Such parsed translation pairs are hard to collect. This paper describes a systemfor finding parsed translation pairs (or translation patterns) that are valid for thetranslation pattern base by comparing wrong translations and corresponding correcttranslations.View full abstractDownload PDF (1354K)
- |<
- <
- 1
- >
- >|