Structured texts typically have a hierarchical structure like trees. Recently, the amount of structured texts have been increased by building large scale digital libraries. These structured texts described by markup language such as SGML (Standard Generalized Markup Language) have not only contents but also structured information.
Previous information retrieval systems cannot manage or retrieve structured information in texts. As structured information becomes available, there is a need to evolve tools to take advantage of structured information. For structured information retrieval, we need new query language which is able to describe structures of texts. Because the structures of texts are expressed by query language only, a query language is the essential element of structured information retrieval. Most of the previous researches have devoted on the representation capability of query language. Because the structures of texts are difficult for end users to recognize and represent it, end users' aspects should be considered in designing a query language.
This thesis present a query language for structured information retrieval. The language expresses the structures of texts by using tree representation. For the language, we made a query system which generates tree structures of texts automatically by parsing DTD(document type definition). The query language enables end users to recognize the structure of texts and to construct a robust query. We apply it to the TDMS(Textcorpus and Dictionary Management System) for the applications.