Representing Documents Using an Explicit Model

Bartell, p. 2 ment is obtained for the artificial MED database of medical abstracts. The generally favorable performance suggests interesting potential for methods which explicitly modify the retrieval system to meet inter-document semantic constraints. 1 Introduction A well known and critical problem for document retrieval systems is the failure of individual keywords to identify the conceptual content of documents. It is difficult to determine whether a document is relevant to a query, or whether two documents are conceptually related, simply by examining the terms they share. In this paper we present a method for representing documents which is based explicitly on the semantic relationships between documents in addition to the individual term usage in documents. This allows documents to be retrieved because they are semantically related, rather than simply related by term co-occurrence. We demonstrate experimentally that this can improve document retrieval performance relative to current popular retrieval algorithms