Full-text Database of Scientific Documents with Various Structures

摘要

A grammar-based full-text database system is presented for handling scientific documents marked up with Standard Generalized Markup Language (SGML). It is necessary for afull-text database system to harmonize the diversity of structures of documents, and therefore, we adopted grammatical methods for transforming them. This is accomplished as follows: (i) documents are stored in original form, (ii) each logical element is assigned to items in a database view, and (iii) when a user issues a query, through lexical and syntactical analysis, logical structures of documents are transformed into a database view and extract information from documents. As a result, our new system provides an information retrieval facility with a uniform view. We expect this method will also enable future implementation of a "text browser" system.