Academic journal article Information Technology and Libraries

Precision and Recall in Title Keyword Searches

Academic journal article Information Technology and Libraries

Precision and Recall in Title Keyword Searches

Article excerpt

This study examines whether adjacency operators in title keyword searches are effective in improving precision. Title keywords from a random sample of titles in economics were searched in FirstSearch with and without adjacency of the keywords specified. Subject headings of the items retrieved were compared with the sample title subject headings to determine the degree of match or relevancy, and figures for precision and recall were calculated. When keywords were discipline-specific, adjacency operators improved precision with little degradation in recall. Systems that allow positional operators or rank output by proximity of terms may increase search success.

Online catalogs provide many opportunities for creative subject access, including keyword searches. While keyword searches in controlled vocabulary fields allow access to subject headings when entry terms or word order are not known, titles also contain subject-rich terms. These keywords use the author's own terminology, which is often more current than the Library of Congress Subject Headings (LCSH),[1] and can be combined or related to each other in order to vary the search. This study investigates the extent to which title keywords convey subject content and compares the relative effectiveness of searching title keywords via two different strategies.

Unlike searches in non-keyword-based systems, which must match the beginning of the field, keyword searches involve identifying the requested terms at any position in the field being searched. Multiple terms can be combined in a search using the Boolean operators AND, OR, and NOT. Word stems or truncated terms, as well as positional operators, can be specified. These operators can specify the order in which the terms appear, their proximity to each other, or that the terms be adjacent to one another. The options in keyword searching allow the user to broaden or narrow a search as needed.

Peters and Kurth determined from a study of dial-access transaction logs at the University of Missouri-Kansas City that library patrons were using title keyword searches as a form of uncontrolled vocabulary search.[2] In other studies, users were observed using title terms for subject access both in the catalog and while browsing the shelf.[3] These studies make a case for the existing use of subject access through title keywords, but show no evidence of the success of these searches, nor the relative success of different types of keyword searches.

Other studies have found title terms used for subject searching: Larson has described the decline of subject searching and the concomitant rise in title keyword searching over a six-year period,[4] and Ensor describes several studies which show a rise in keyword searching of all types.[5] Both authors note that keyword use rises with catalog experience. Connell observed that experienced users perform title keyword searches as a lead-in to the controlled vocabulary,[6] and Peters and Kurth recommend this method in addition to using title keyword searches alone.[7]

When users perform title keyword searches as a subject approach to the catalog, how good are the results? More specifically, do items that contain the same terms in their titles cover the same topic, and are certain title keyword search strategies more effective than others for subject searching?

Literature Review

Characteristics of Title Keyword Searches

Title keyword searching has some advantages over controlled vocabulary subject access. Titles terms are more likely to agree with the user's terminology and serve as a complement to the assigned subject headings[8] and have been found by Jamieson to overlap very little with subject cross-references.[9] Bates found that subject experts in economics consistently preferred headings that were more precise than the subject headings assigned to works, and that they particularly disliked the subheading "economic conditions" because of the variety of meanings covered by it. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.