Query Processing in Context - Oriented Retrieval of Information

Abstract

This paper proposes a context-oriented approach towards document retrieval and presents the query processing component of the retrieval process in detail. Our approach differs from the existing ones by its use of word sense disambiguation in the first stages and by its employment of query structuring. The proposed system, called CORE, attempts to match contexts of documents and queries rather than regarding these as sets of words. Clarifying the indicated meaning of the words in the query before the search is conducted will eliminate a large number of irrelevant documents and will enable the construction of a context for the query. Query structuring allows the preparation of a query execution plan for an efficient search and thus yields response time improvement. In addition, it positively affects Recall and Precision by considering the relative importance of the words in the query.