A pilot study on the relative retrieval effectiveness of semantic relevance (by terms) and pragmatic relevance (by citations) is reported. A single database has been constructed to provide access by both descriptors and cited references. For each question from a set of queries, two equivalent sets were retrieved. Ail retrieved items were evaluated by subject experts for relevance to their originating queries. We conclude that there are essentially two types of relevance at work resulting in two different sets of documents. Using both search methods to create a union set is likely to increase recall. Those few retrieved by the intersection of the two methods tend to result in higher precision. Suggestions are made to develop a front-end system to display the overlapping items for higher precision and to manipulate and rank the union sets retrieved by the two search modes for improved output.The study of relative retrieval effectiveness using different document representations is not new. Over the last two decades, many researchers have tested the performance of various document representations.They include the use of title words, terms in the abstracts, free-text terms, terms from controlled vocabularies, and any combination of these. Katzer and his colleagues at Syracuse University succinctly reviewed and summarized the literature [ 11. They noted that most studies have been based on exceedingly small databases and many have used fewer than ten documents or queries. These studies often examined the retrieval from the use of more than one representation without regard of whether the retrieval was relevant to the originating query. Their own study of seven different representations showed that different document representa-*This publication was supported in part by NIH grants ROl-LM-04177 and K04-LM-00078 from the National Library of Medicine.Received March 16,1987; revised April 21, 1987; accepted April 27, 1987. 0 1989 by John Wiley & Sons, Inc.tions tended to retrieve different sets of documents. None was able to achieve a recall ratio of more than 41% and a precision of over 43%. These results generally corroborated those found by others. This group of studies was confined to comparisons of retrieval results using document representations based on terms found in the bibliographic record, that is, representations derived from semantic relevance alone. The relevance relationship between the semantics or meaning of the terms and content of the document underlies this type of document representations.A second group of studies includes aspects of citation relations. They were not strictly comparisons of retrieval results. Salton showed that there was a positive correlation between papers with citation similarities and those with index term similarities [2]. Papers with stronger bibliographic linkages and those co-cited in many papers also shared many more assigned descriptors [3,4]. Several other studies demonstrated that citations were useful feedback data in retrieving similar articles [5,6]. These early studi...
No abstract
Much work has been completed on the application of quantitative measures to literatures dealing with the subject area devoted to ideas or concepts. Little or no work has been reported on the use of these techniques in an applied situation, on a literature devoted to a scientific product. This study attempts to apply the quantitative measures to such a literature and examine some of the implications of the results in terms of professional action on the part of the librarian or information scientist. Bibliometric techniques developed on long-lived literatures are applicable to short-lived literatures devoted to a sample of two pharmaceutical products.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2025 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.