Metadata, recall, and abstracts: can abstracts ever be reliable indicators of document value?

Abstract

Abstracts from seven Internet subject trees (Euroferret, Excite, Infoseek, Lycos Top 5%, Magellan, WebCrawler, Yahoo!), five Internet subject gateways (ADAM, EEVL, NetFirst, OMNI, SOSIG), and three online databases (ERIC, ISI, LISA) were examined for their subject content, treatment of various enriching features, physical properties such as overall length, and their readability. Considerable differences were measured, and consistent similarities among abstracts from each type of source were demonstrated. Internet subject tree abstracts were generally the shortest, and online database abstracts the longest. Subject tree and online database abstracts were the most informative, but the level of coverage of document features such as tables, bibliographies, and geographical constraints was disappointingly poor. On balance, the Internet gateways appeared to be providing the most satisfactory abstracts. The authors discuss the continuing rle in networked information retrieval of abstracts and their functional analogues such as metadata.

Citation

Wheatley, A. and Armstrong, C.J. (1997), "Metadata, recall, and abstracts: can abstracts ever be reliable indicators of document value?", Aslib Proceedings, Vol. 49 No. 8, pp. 206-213. https://doi.org/10.1108/eb051467