Journal of the Association for Information Science



Bert R. Boyce




Jean Tague-Sutcliffe, 1931---1996
Mike Nelson

With sadness we begin with a memorium to Jean Tague-Sutcliffe, last year's winner of the ASIS Best Information Science Book Award and a co-author of an article in this issue.




Design and Implementation of Automatic Indexing for Information Retrieval with Arabic Documents
Ismail Hmeidi, Ghassan Kanaan, and Martha Evens

 Hmeidi and Evens study Arabic citations and abstracts on computational linguistics, with queries and relevance judgments for each query provided by computer science students. Terms occurring in the abstracts 2 or more times, but not more than 260 times, were chosen as index terms using collection frequency of all words, then all roots and finally all stems. In Arabic the token to type ratio is much lower than in English and the inverse document frequency ratio is typically much higher. At high recall levels the root and stem indices outperform the word index. Automatically assigned terms outperformed manual assignment consistently for words, and at higher recall levels for both words and stems.




Information Using Likeness Measures
Martin Frické 

Frické defines information in terms of possible worlds constructed by choosing individual items of interest and traits that may or may not be present in these items. If we know the number of possible worlds then the amount of information in a proposition is measured by the number of possible worlds it excludes, and the difference between propositions depends upon how much they disagree in their basic states of items assigned to traits.




Types and Levels of Collaboration in Interdisciplinary Research in the Sciences
Jian Qin, F. W. Lancaster, and Bryce Allen

Using the 1992 edition of Science Citation Index, Qin, Lancaster, and Allen select a sample of 1000 articles randomly and reduce it to 846 by elimination of those with over 15 authors or with no references. The journal titles of the references to each article are then grouped by Ulrich's subject categories excluding comprehensive journals. The number of categories present then measured the degree of interdisciplinarity. Each article also received a level code for no collaboration, in department collaboration, in the same institution collaboration, multiple institution in a country, or finally international collaboration. From the collaborative articles with at least two disciplines, fifty first authors were surveyed. Collaboration correlates with interdisciplinarity but not in all disciplines. Institution type, nature of problems, personal contact, and funding all affect collaboration. Interactive high speed mechanisms are preferred for communication. There is considerable variation in level of collaboration among disciplines.




Measuring the Impact of Information on Development: A LISREL-Based Study of Small Businesses in Shanghai
Liwen Qiu Vaughan and Jean Tague-Sutcliffe
Variables for information and other factors impacting economic development were identified by Vaughan and Tague-Sutcliffe in discussions with experts and users. In the pilot variables were determined from a sample of small businesses. The study issued 3000 questionnaires, and 376 were returned. After an additional 100 questionnaires and some deletions 450 data points remained. An initial run showed that the data did not fit the original model. The addition of a new latent variable for development and the removal of the latent variable for expertise resulted in a good fit. Business environment and information use both make significant positive contributions to success. Informal information sources make a stronger contribution.




Clustering and Classification of Large Document Bases in a Parallel Environment
Anthony S. Ruocco and Ophir Frieder 

Employing a single pass algorithm with the cosine coefficient, Ruocco and Frieder use a parallel processing environment to cluster a TIPSTER subset. Clusters are assigned to processors so that a document can be simultaneously compared to as many clusters as there are processors. A single link classification process was also performed. Significant performance time reductions are possible using commercially available equipment, and the need for large pre-computed matrices for single link structures can be avoided by computing only values as they are needed.




 Fractional Counting of Multiauthored Publications: Consequences for the Impact of Authors
G. Van Hooydonk

 In the first of three brief communications related to citation analysis in this issue, Van Hooydonk proposes a proportional ranking where the author's relative position in the citation R is used to obtain a relative weight, (N + 1 - R)/(N(N + 1)/2). Not only number of credited publications but impact measures are strongly effected by the method of computation used.




Structural Modeling of Network Systems in Citation Analysis
Dang Yaru

Building on standard citation analysis procedures Yaru develops a graph theoretic model that may shed some light on the development of topics in the structure produced.




The Diffusion of Scientific Journals Analyzed through Citations
Pedro Alvarez and Antonio Pulgarín 

Alvarez and Pulgarín suggest a variable ``diffusion'' defined by the number of citations received in a year by a journal. Using the Rasch model, the probability that a particular journal has been cited in a particular year can be computed. The Rasch diffusion measure using 10 years of Physics journal citation counts from the 1994 SCI Journal Citation Reports ranks the journals differently than would impact factor.




Information Systems Development and Data Modeling: Conceptual and Philosophical Foundations
by Rudy Hirschheim, Heinz K. Klein, and Kalle Lyytinen
reviewed by: Peter Aiken




Symbolic Projection for Image Information Retrieval and Spatial Reasoning
by Shi-Kuo Chang and Erland Jungert
reviewed by: Geoffrey Z. Liu 





ASIS HomeSearch ASISMake A Comment

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