Discusses methods for automatically compiling domain-oriented vocabularies in natural language systems and describes techniques for evaluating the quality of the resulting word lists. A study is described that used subject headings from Grolier's Encyclopedia and the United Press International newswire, and filters for removing high frequency…

The current state of information retrieval (IR) evaluation is reviewed with criticisms directed at the available test collections and the research and evaluation methodologies used, including precision and recall rates for online searches and laboratory tests not including real users. Automatic text retrieval systems are also discussed. (32…

These two articles discuss clustering structure in the Cystic Fibrosis Document Collection, which is derived from the National Library of Medicine's MEDLINE file. The exhaustivity of four subject representations and two citation representations is examined, and descriptor-weight thresholds and similarity thresholds are used to compute…

Examines the use of various statistical techniques for generating domain-oriented multiword vocabulary terms for natural language database systems. Conclusions show the vocabulary clustering effect should be considered when making significance calculations and that a simple ratio of subject matter relative frequency to total sample relative…