Representing Documents Using an Explicit Model of Their Similarities

A method is proposed for creating vector space representations of documents based on modeling target inter-document similarity values. The target similarity values are assumed to capture semantic relationships, or associations, between the documents. The vector representations are chosen so that the inner product similarities between document vector pairs closely match their target inter-document similarities. The method is closely related to the Latent Semantic Indexing approach; in fact, they are equivalent when the target similarities are derived directly from document similarities based on term co-occurrence. However, our method allows for external sources of inter-document semantic constraints to be used in the indexing, though at greater computational expense. The method is applied to three standard text databases from the information retrieval literature. On the CISI database of information science abstracts, performance (measured by precision averaged over a range of recall levels) improves by 28% compared to a weighted term-vector approach, and improves 10% compared to Latent Semantic Indexing. Similar improvement is obtained on the Cranneld database, but no improve-1 Bartell, p. 2 ment is obtained for the artiicial 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.