LRTS v 40, no. 2, April 1996

Handling Spelling Errors in Online Catalog Searches

Karen M. Drabenstott and Marjorie S. Weller

The purpose of this paper is to add to our understanding and knowledge of spelling errors in online catalog searches based on empirical studies of spelling errors in online catalog searches and suggest ways in which systems that detect such errors should handle the errors that they detect. One study of spelling errors in online catalog involved a categorization of user queries for subjects that were extracted from four university libraries' online catalog transaction logs. The results of the analysis demonstrated that less than 6% of user queries that match the catalog's controlled and free-text terms contain spelling errors. This percentage did not account for spelling errors in user queries that failed to match the catalog's controlled and free-text terms, because of the difficulty of verifying certain terms and phrases and of collection failure. The results of a related study involved user responses to an experimental online catalog that detected possibly misspelled words• While the majority of users corrected misspelled query words, a sizable proportion made an action that was even more detrimental than the original misspelling; for example, they added another ward or phrase to the query in addition to the misspelled word. This paper concludes with three recommendations for improvements to online catalogs to assist users in the correction of misspelled query words and the detection of queries that fail due to collection failure.

Tables of Contents in Library Catalogs: A Quantitative Examination of Analytic Catalogs

Claus Poulsen

Easy access to tables of contents from vendors and the technological development of optical character reading have actualized access to articles in books via tables of contents in library catalogs. From earlier studies we know that analytic book catalogs can provide access to up to 600% more works than the traditional catalog by simply adding analytics for works in composite works to the catalog. In this study we examine the proportion of composite works and the number of articles in these books in two different university libraries. The influences of library type, publication language, subject field, and date of publication are examined, and the results are compared to previous studies. The proportion of composite works is between 10% and 20%. The number of articles in the composite works varies from 20 to 30 articles per book -- highest for the sciences and the English-language publications and lowest for the social sciences.

Reshelving Study of Review Literature in the Physical Sciences

Nancy J. Butkovich

Review publications contain articles that give overviews or state-of-the-art reports on specific topics. Although some review titles are published more frequently, many appear only once a year. At the Physical Sciences Library of the Pennsylvania State University's University Park Campus, a year-long reshelving study of the review publications collection was undertaken to determine usage of the titles. This need was fueled by a lack of shelf space, storage considerations, and the threat of serials cancellations. Three hundred review titles were examined. The best data were found in classes QC, QD, and QII-QP (monographic series only). The other classes had few titles or low use tallies. Approximately half of all titles were used at least once. Periodicals had a higher percentage of use than did monographic series.

The Structure of the Library Market for Scientific Journals: The Case of Chemistry

Stephen J. Bensman

In this paper, the author analyzes the skewed distributions of price and scientific value that constitute the structure of the library market for scientific journals, using chemistry as a test case. A numerical index constructed from a survey of Louisiana State University chemistry faculty and total citations taken from the Science Citation Index Journal Citation Reports were utilized as measures of scientific value. Methodological problems arise from the skewed distributions customary in library research. The major findings are (1) that scientific value does not play a role in the pricing of scientific journals and (2) that little relationship consequently exists between scientific value and the prices charged libraries for scientific journals. Libraries have the opportunity to implement a massive restructuring of their serials collections. A software package named the Serials Evaluator is described. Under development at Louisiana State University, it is software for the automated selection of journals for cancellation and remote access through document delivery.