This thesis describes a fully automated indexing system for an information retrieval system. In information retrieval, indexing is the task consisting of the assignment to-stored records and incoming information requests of the content identifiers capable of representing records or query contents.
The indexing system described in this thesis is performed automatically excluding any manual labour. The procedures necessary to implement this automatic indexing system are laxical analysis, stop-list construction, thesaurus construction.
Initially, the dictionaries are constructed in the form of the ISAM file structure using the selected index terms. Afterwards, when the updating of the dictionaries is necessary, the dictionary can be enlarged by adding the supplementary index terms automatically.
Since the available main memory to a user program may be limited for implementation of the automatic indexing system described in this thesis, the technique of chaining is used. The performances of the semi-automatic indexing and the fully automatic indexing system proposed in this thesis are compared. The steps of the enlargement of the dictionary are also shown. As the document abstracts are processed the size of the dictionary is gradually enlarged automatically. The graph of the documants size vs. the dictionary size is shown in the appendix.
정보 감색 과정에 있어서 제일 중요한 부분인 indexing 과정에서 종래의 사람의 노력이 필요했던 반자동 indexing 과정을 program이 자동적으로 처리해 주는 완전 자동인 indexing system을 만들었으며, 이 system이 작동하기 위한 여러가지의 필요한 database 들을 구성하고 그들의 efficiency와 이론적인 배경, 그리고 indexing 이론들을 설명하였다.
이 논문에서 만든 완전 자동 indexing system의 실험결과를 분석하고 능률을 검토한 결과 만족한 결과를 얻었고, system에 의한 dictionary들의 자동적인 확장이 보여졌으며 따라서 본 논문에서 만든 자동적 indexing system의 타당성이 증명되었다.