A method of retrieving documents, based on domain-specific context and meaning, is described in this paper. This is a major and significant improvement over keyword-based search methods, including those employing keyword expansion.

The process for developing a domain-specific ontology (an abstract view of a domain, through a collection of concepts and their interrelationships) is well laid out, and thoroughly described. Potential complications, such as inconsistencies in the ontology, are discussed, and solutions provided. Problems with traditional methods of retrieval are outlined, and comprehensive references to the literature on retrieval methodology, keyword expansion, and ontology-based search are provided.

In the discussion of user queries, procedures for generating tokens from the text of the query, pruning and disambiguating keywords that match more than one query, and query expansion and structured query language (SQL) query generation, are thoroughly described. Algorithms and/or formulas are presented for each of these steps.

Finally, the results of empirical trials and future extensions are considered. The ontology-based model performed at least as well as, or better than, keyword-based methods in a variety of circumstances, as measured by both recall and precision. The conclusions presented by the authors are sound and reasonable, with good suggestions for further work.

This is an excellent paper, which makes an important contribution to the literature on information retrieval. The presentation is clear and complete. The paper will be of interest to researchers in the field of information retrieval, and will be particularly useful for those involved in the retrieval of multimedia (nontextual) data.