1996 Fiscal Year Final Research Report Summary
Study of Intelligent Retrieval Function for Engilish Corpus
| Project/Area Number |
07680425
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報システム学(含情報図書館学)
|
|---|
| Research Institution | Fukui University |
|---|
Principal Investigator |
OGURA Hisakazu Dept.of Comp.Sci., Faculty of Eng., Fukui Univ., Prof., 工学部, 教授 (50127940)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
ODAKA Tomohiro Dept.of Comp.Sci., Faculty of Eng., Fukui Univ., Assoc.Prof., 工学部, 助教授 (10224247)
|
|---|
| Project Period (FY) |
1995 – 1996
|
| Keywords | English Corpus / Intelligent Retrieval System / Natural Language Processing / Whole Text Database / Grammatical Function / Patial Parsing / Bottom-up Parsing / CESP |
|---|
| Research Abstract |
The airms of this study is to try to apply the parsing method to retrieve words in english copus and to improve retrieval fitting accuracy.
We execute retrieval of words or phrases for a variety of purposes making use of retrieval systems. A word in a sentence, generally, has the function (s) which can be defined in relation to other elements of the sentence. In this paper we will refer to this sort of function as grammatical function of word. And sometimes it is the case that a word can fulfilll different functions in different sentences. Thus the word "there" is used either as a deictic locative adverb or as an expletive element. The problem for a retrieval system is the following : when we are in search of expletive "there" sentences, for example, the sentences containing locative "there" are irrelevant. It is necessary to decrease the number of these unexpected sentences in the result of retrieval, possibly up to zero. We have attempted to incorporate this feature into the retrieval system for the database called English Corpus by making use of grammatical function of word. This paper is a preliminary discussion of a prototype retrieval system applied to the English textbooks used at Japanese junior high schools.
We acomplished initially a retrieval system by kcl, a LISP system, and transfered and implimented to CESP system which is a prolog like language with object oriented functions. The rules in the system are rewritten and reconstructed and make the rule-base separately. We tried a bottom-up pasing in order to extract a partial parse tree which includes the word to be retrieved.
The research is still in progress.
|
