Management and Design Tools for Document Retrieval Systems

Design and management of Document Retrieval Systems (DRSs) requires tools and performance measures to aid in selecting preferred options and in controlling the fundamental processes of inquiry analysis, indexing, retrieval, and system growth. This study presents a step toward the generation of such tools by developing a retrieval quantity estimate. Defined as a function of the inquiry form, search strategy, and descriptor-document distribution, the estimate can be used to predict quantity output of inquiry, measure impact on quantity output due to system growth, and aid in the tuning of the indexing and formal inquiry specification processes. The definition of the retrieval quantity measure is based on the identification of certain canonical forms that characterize the underlying principles of DRS indexing and retrieval. Tested on an operational DRS, the estimate demonstrated high prediction accuracy for a variety of typical inquires. The methodology for determining retrieval quantity appears to hold for a very wide range of system size, subject content, and construction. 225 pp. Bibliog.