Retrieval Effects of Query Expansion on a Feedback Document Retrieval System

摘要

This thesis presents the results of an experimental investigation into the effects that various forms of query expansion have on the retrieval effectiveness of a document retrieval system (DRS). A comprehensive theoretical overview of the probabilistic approach to modelling a DRS is given, and from this theoretical background some heuristic query expansion techniques are proposed. Retrieval results are given when query expansion strategies using the maximum spanning tree, nearest neighbors to query terms, or index terms from known relevant documents, are incorporated into an iterative feedback retrieval system.