Analysis of Questions Addressed to a Medical Reference Retrieval System: Comparison of Question and System Terminologies.

Requests for subject and author searches submitted to the Medical Documentation Service of the Library of the College of Physicians of Philadelphia were studied. A total of 483 subject reference questions were analyzed for the number of question terms matching system subject headings (M), number of question terms translatable to system subject headings (1), number of stop-list words (5), and number of untranslatable words (U), using the judgment of the author. The average question had one M term or 22% M, one T term or 21 % T, two S words or 38% S, and one U word or 19% U. Thus 81 % of the average question was accounted for by M+T+S; in addition, 46% of the questions had no U words. Analysis of variance failed to show significant (5% level) differences between doctors' and lawyers' questions or between negotiated and nonnegotiated questions in number of M, T, S, or U. Study of limitations on searches for the 483 subject search requests and 38 requests for author searches showed that the requestor rarely stipulated type of material to be covered, languages to be included, time period to be covered, cost, or time for completion. [ABSTRACT FROM AUTHOR]