1999 Fiscal Year Final Research Report Summary
Research on Structural Knowledge Acquisition for Japanese Expressions from Large Corpus
| Project/Area Number |
09558035
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 展開研究 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Tottori University |
|---|
Principal Investigator |
IKEHARA Satoru Tottori University, Faculty of Engineering, Professor, 工学部, 教授 (70283968)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
SHIRAI Satoshi ATR Interpreting Telecommunication Laboratories, Head of Department 3, 第3研究室, 室長
|
|---|
| Project Period (FY) |
1997 – 1999
|
| Keywords | n-gram / collocation / Noun Phrase / Sentence Structure / Complex Sentence / Dependency Relation / Retrieval by Meanings / Semantic Dictionary |
|---|
| Research Abstract |
The purpose of this theme is to establish methods to obtain the knowledge about Japanese sentences from large corpus and to develop the program tools for them.
First, the method to automatically acquire dependency rules for Japanese noun phrases from large corpus were investigated. In this study, we proposed the new method based on the semantic attribute system for Japanese nouns. It was confirmed that the dependency rule obtained by this method can analyze noun phrases with the accuracy of 88%.
Second, applications of n-gram statistical method to find collocational expression from large corpus were investigated. In this study, we proposed rewriting methods of original text in order to close up the underlined expressions. This method was developed aiming at gathering the patterns of complex sentence and noun phrases.
The third is related to algorithms for retrieval of sentences or expressions based on meanings and dependency structure. These algorithms are used for example based machine translations and sentence pattern dictionary developments.
|
