Alternative representation for documents and queries

It is commonly thought that there are but two choices of a text representation scheme for use in an information retrieval system - the free-text, natural language representation of documents' contents, or a controlled vocabulary representation based on a thesaurus developed for a particular topic area. The research herein reported suggests that there is a third alternative - one which combines the advantages of both traditional approaches and which overcomes many of the difficulties associated with each. The representation scheme has already produced promising preliminary results. As an approach to real-time document representation, its performance suggests a truly viable alternative for document representation.