On the Use of Clustered File Organization in Information Search and Retrieval

Abstract

In modern retrieval environments, collection searches are normally conducted on-line under user control. Iterative collection searches can then be performed where tentative queries are initially processed, to be successively improved and refined during the search process. When searches are carried out directly by the users, classified document organizations are especially useful, because collection browsing becomes possible, and access is provided to complete groups of related documents. The state-of-the-art in automatic document classification is summarized in the present study, and evaluation data are provided to demonstrate the effectiveness and efficiency of clustered file search operations.