Automatic Dictionary Construction; Part II of Scientific Report No. ISR-18, Information Storage and Retrieval...

Part Two of the eighteenth report on Salton's Magical Automatic Retriever of Texts (SMART) project is composed of three papers: The first: "The Effect of Common Words and Synonyms on Retrieval Performance" by D. Bergmark discloses that removal of common words from the query and document vectors significantly increases precision and that synonyms were more effective for recall than common words. Paper two: "Negative Dictionaries" by K. Bonwich and J. Aste-Tonsmann discusses a rationale for constructing negative dictionaries and examines the retrieval results of experimentally produced dictionaries. The third paper: "Experiments in Automatic Thesaurus Construction for Information Retrieval" by G. Salton describes several new methods for automatic, or semi-automatic, dictionary construction, including procedures for the automatic identification of common words, and novel automatic grouping methods. The resulting dictionaries are evaluated in an information retrieval environment. (For the entire SMART project report see LI 002 719, for Part One see LI 002 720 and for Parts 3-5 see LI 002 722 through LI 002 724.) (NH)