Journal of the Association for Information Science



Bert R. Boyce




Jack Belzer, January 16, 1910 - June 22, 1995
Anthony Debons




Correspondence Factor Analysis of the Publication Patterns of 48 Countries over the Period 1981-1992
Jean-Christophe Doré, Tiiu Ojasoo, Yoshiko Okubo, Thomas Durand, Gérard Dudognon, and Jean-François Miquel

Doré describes the learned publication output of the 48 most prolific countries in the Institute for Scientific Information database from 1981-1992. Data reduction is completed using Correspondence Factor Analysis which creates maps similar to those produced by multidimensional scaling. Countries are related to one another and to 17 disciplinary areas identified by ISI. Unlike Principle Components Analysis the distances are not Euclidean but Chi Square metrics giving the probability of a statistical link between countries and disciplines.




Multiple Search Sessions Model of End-User Behavior: An Exploratory Study
Amanda Spink

Multiple end-user searches appear to be the norm with needs and strategies changing over time. Nearly 80% of a sample of users interviewed by Spink were in the first stage of their information-seeking behavior. Over half of the multiple search users reported they were in stage one of a five-stage scale toward completion, 6% in stage two, and 40% in one of the last three stages. Multiple searches over the life of the project are common. Search terms and strategies had been modified by 70% of the multiple search users.




Simulation Model for Journal Subscription by Libraries
Richard E. Quandt

Quandt's model assumes that libraries faced with rising journal costs will cut subscriptions, and publishers faced with falling subscriptions will raise prices. Not surprisingly it predicts large falls in subscriptions before[jyany stability is reached. Since cooperative subscription schemes, and non-subscription revenue sources are not modeled, and the actual cancellation policies likely to be in use are unclear, the value of this model seems theoretical.




Interlinking E-Mail Systems
Dale Hibner, Richard Bellaver, and Frank M. Groom

Hibner discusses, at a management rather than technical level, facilities for connecting local e-mail systems together to serve a whole widely dispersed organization. A global system must acquire name and location information from local systems on a continuous basis and translate it into the global format prior to updating its directory. A global system may simply match and forward messages, or it may maintain its own secure modem pool and a complete mailbox system. A gateway design connecting separate systems is also possible.




Reading the Bones: Information Content, Value, and Ownership Issues Raised by the Native American Graves Protection and Repatriation Act
Teresa Olwick Grose

The informational aspects of human remains and artifacts provide a unique viewpoint for an intriguing contribution by Grose. The Native American Graves Protection and Repatriation Act considers the items it protects valuable because of the information they convey about the culture of their origin, and implies that the original owners of their descendants identify which objects are to be within the scope of the law. Provenance of the objects instills most of their religious and symbolic value just as it provides most of its scientific value to the archaeologist. But both the provenance and the object are necessary; the medium and its messages are inseparably linked. Field notes and other records may thus be included in the scope of the act. Ownership of scientific data and the right to free exchange of information here seem conflicted with personal privacy and confidentiality.




Method for Evaluation of Stemming Algorithms Based on Error Counting
Chris D. Paice 

Two articles address quite different aspects of the retrieval problem. Paice develops measures to evaluate a stemming algorithm based upon ideal groupings prepared manually. The indices developed, while they very considerably from one word source to another, seem to produce consistent rankings. It is possible to distinguish among heavy and light algorithms, where a light stemmer will make few overstemming errors but leave many understemming errors in place, and a heavy stemmer will attempt to go forward even with a high risk of overstemming and thus conflate terms with distinct meanings.




Levels of Citation Relations between Papers
P. Pichappan 

Using terminology from Ranganathan, Pichappan suggests a four-type classification scheme for citations. Type 1 relationships, require both cited and citing papers to be basically similar. Type 2 relationships imply that the cited paper forms the body of the discussion; type 3 that the whole of the cited paper relates to part of the citing paper. Type 4 relationships presuppose a similarity between small parts of both the cited and citing paper. Assuming that proper evaluations could be made as to the class of each citation relationship, it is speculated that the use of only type 1 citation relationships to compute document document similarity classes would result in the clustering of highly similar documents.




Telecommunications Politics: Ownership and Control of the Information Highway in Developing Countries
edited by Bella Mody, Johannes M. Bauer, and Joseph D. Staubhaar
Reviewed by Alice Robbin




The Impact of Emerging Technologies on Reference Service and Bibliographic Instruction
edited by Gary M. Pitkin
Reviewed by Donald R. Smith




Human Factors in Information Systems: Emerging Theoretical Bases
edited by Jane M. Carey
Reviewed by Thomas A. Peters





ASIS HomeSearch ASISMake A Comment

© 1998 , Association for Information Science
Last update: November 06, 1998