Search

The author's journal reference cards for 1965 and 1966 were analysed according to three "interest patterns": (I) the total collection of 1469 article titles, a "potentially useful" set; (II) a subset concerning only his research specialty (deoxyribonucleic acid, histone, chromosomes, or radiation); and (III) a subset of articles defined as "useful" (those actually read by March, 1967). For each pattern, journals were ranked by frequency of use and a scatter diagram was drawn. Patterns I and II largely resembled patterns obtained by counting citations in basic journals or by counting publications of selected researchers. Pattern III was more widely scattered. It is concluded that access to diverse journals is needed by researchers to supply new ideas, and that this diversity of reading is not reflected adequately by citation counting or other indirect means. Current Contents, used by the author for current-awareness purposes in building his card file, generated 88% of all articles. Scatter diagrams indicated the decreased scatter predicted from its use. The 30 most important journals in this collection, including about 80% of titles, are ranked for each pattern. [ABSTRACT FROM AUTHOR]

The paper is a report of most of the major findings of a study in searching the periodical eye research literature. Questions were collected from eye researchers and a selected group of these were searched in nine different secondary sources. Articles thought to be relevant were Xeroxed and sent to the eye researchers who subsequently rated the articles. Articles of eye research interest are found in a wide variety of journals, but a small number of journals carry a large proportion of the articles judged valuable by the eye researchers. Approximately a fourth of eye research articles are in foreign languages. Translations ore not readily available. Despite a delay of more than 15 months between the original appearance of articles in journals and the mailing of photocopies, about half of the articles of interest to the researchers were not known to them previously. For extensive retrospective searches more than one secondary service must be used. Index Medicus and Excerpta Madica (Section 1 2) or Ophthalmic Literature would be good sources. MEDLARS demand searches were not shown to be clearly superior to manual searches of index Medicus. Titles, abstracts, and full text were shown to be equally effective in permitting searchers to retrieve references that were subsequently rated as relevant by the researchers. A searcher with a background in ophthalmology was able to retrieve more articles of research interest than other non-ophthalmologist searchers. [ABSTRACT FROM AUTHOR]

The recent evaluation of Medlars has shown that it is possible to undertake an investigation to measure the effectiveness of a large mechanized retrieval system and to identify the factors that are most important in controlling and limiting the effectiveness of the system. A one-time evaluation, however, can only measure the performance of a system at a particular point in time. When changes are made to the system, as a result of the investigation, we would like to be able to estimate the effects of those changes. Moreover, a one-time evaluation, however comprehensive, cannot identify all sources of system failure. It can locate the most important sources and the areas of greatest weakness, but it cannot identify, for example, all specific instances of vocabulary inadequacy in the system.

A comprehensive program to evaluate the performance of MEDLARS was conducted by the National Library of Medicine in 1966 and 1967. This report describes the methodology used and presents a summary of the principal results, conclusions, and recommendations. [ABSTRACT FROM AUTHOR]