Using syntactic analysis in a document retrieval system that uses signature files

Abstract

Our work involves the study of the extent to which natural language processing techniques aid the automatic indexing and retrieval of documents. In this paper we describe the use of signature files in large text retrieval systems. We show that good performance can be obtained without requiring the significant overheads required for the inverted file technique. We examine the use of syntactic analysis of the text in all stages of retrieval and argue that an initial Boolean query should be performed that provides a subset of documents, which are then ranked. We then give an algorithm for generating such queries, taking into account the syntactic structure of the queries.