The use of automatically-obtained keyword classifications for information retrieval

This report summarizes the work carried out on the automatic construction of keyword classifications and their use in information retrieval that we have carried out in the last eighteen months. It discusses the possible characteristics of such classifications, and potential ways of using them; and describes the various approaches to classification we have considered, in terms of an overall frame of reference in which several types of classification are distinguished. The results of experiments using a collection of documents and test requests obtained from the Aslib Cranfield project are presented, from which it can be concluded that automatic classifications which lead to noticeable improvements in retrieval performance over those obtainable with unclassified terms can be set up.