Conceptual Information Retrieval System Based on Automatically Constructed Semantic Word Network

Abstract

This chapter describes a method to dynamically generate a semantic word network which can properly reflect the semantic status. A conceptual information retrieval system based on the method is also described. The method started with the determination of keywords in a textbase. Keywords can then be clustered using analysis methods. For each cluster of keywords, the hierarchical and sibling interrelationships then are statistically inferred, so that the semantic word network can be dynamically established. The method can be formed as a module in concept information retrieval (IR) system. The query expansion function can be implemented with the module, so that the recall and precision of query result can be improved.