Logic framework for information retrieval

Abstract

Information retrieval systems have been moving towards many of the functions of expert systems. In light of recent developments it is interesting to consider what an ‘expert’ system in information retrieval might be expected to provide. The various models that are currently used to describe the retrieval process include the probalistic model, the Boolean and extended Boolean models, and the vector space model. Each of these models is based on keyword retrieval which operates at a symbolic, text-matching level, and ignores any semantic and contextual information in the retrieval process. It is questionable whether extensions of the traditional approaches to information retrieval will be able to provide the mechanisms needed for more intelligent or more ‘expert’ retrieval systems. Consequently, an alternative view is considered in which retrieval is based on an information structure called a concept space. Finally, a logic framework is presented to define a semantic model, which includes the current retrieval models, for using the knowledge contained in such a concept space. A consistent retrieval framework or theory allows us to formalize the semantics of bibliographic retrieval needed to provide the functional requirements of an expert system. © 1989 John Wiley & Sons, Inc.