B U L L E T I N
Selected Abstracts from JASIS&T
Editor's note: We invite JASIS&T authors to submit structured abstracts of their articles for possible inclusion in the Bulletin, particularly those that might be of interest to practitioners. ASIS&T would welcome reader feedback on the usefulness of this (or any other) Bulletin feature (firstname.lastname@example.org).
From JASIS&T v. 55 (4)
Macpherson, K. J. (2004). An information processing model of undergraduate electronic database information retrieval, (333-347).
Study and Results: In this study, information processing theory is applied to the development of a two-stage model that represents end-user information retrieval from electronic databases. This model addresses a number of issues not covered in previous models that have been put forward in the literature. In a large (n=254) experimental study at the University of Canberra, Australia, concept-based teaching techniques for information retrieval were developed to test the model's predictive utility. Statistically significant results in terms of improvement in knowledge of the search process, and in the ability to locate credible literature, support the model's efficacy as a construct representing the cognitive underpinnings of information retrieval and the effectiveness of the concept-based teaching of information retrieval.
What's New? Effective information retrieval is a key educational outcome in an information society. Theory building in the area and practical teaching strategies derived from this are therefore timely and important. This study provides both a new perspective on theory in information retrieval and demonstrates teaching strategies that can provide significant retrieval improvement to undergraduate novice end-users of electronic databases.
Limitations: Although the sample was large, replication of results in another higher (or other) education setting would validate or refute findings.
Kabel, S., de Hoog, R., Wielinga, B.J., & Anjewierden, A. (2004). The added value of task and ontology-based markup for information retrieval, (348-362).
Study and Results: The added value of ontology-based indexing for information retrieval is investigated. Different index types, varying from flat lists of keywords to structured domain- and task-related
indexing ontologies, are compared in an experiment that simulates the task of creating instructional material from a database of source material. The results show that ontology-based indexing and task-related markup improve the
quality of the created product only to some extent, but that it certainly improves the efficiency and effectiveness of search and retrieval and precision of use.
Limitations: The experiment to measure the added value of ontology-based indexing was conducted in a controlled setting using a relatively small database.
From JASIS&T v. 55 (5)
Zeng, M.L., & Chan, L.M. (2004). Trends and issues in establishing interoperability among knowledge organization systems, (377-395).
Study and Results: In the networked environment cross-domain, cross-subject and cross-language searching are common practices. However, users are often unaware of the diverse knowledge organization systems (KOS) that impede optimal retrieval. These include thesauri, subject headings lists, classification systems and other categorization schemes used to index or organize different databases. This paper surveys activities and research projects aimed at achieving interoperability among KOS and analyzes the methods used. In all, 18 projects were examined and evaluated. The appendix provides a brief description of each.
What's New? Eight conventional and new methods that have proven to be widely accepted and promising were summarized. The article should be of interest to the following groups of information professionals: (1) those who are engaged in or contemplating the initiation of activities in linking or mapping KOS; (2) those who are designing KOS for a wide audience; (3) those who assist users in information retrieval across different domains, languages and communities that use different KOS.
Limitations: The paper is based on published reports and available sources on the Web. Similar projects that have not been published or are not known to the authors are not included in the paper. Website addresses of the projects are also subject to change.
Vinkler, P. (2004). Characterization of the impact of sets of scientific papers: The Garfield (Impact) Factor, (431-435).
Study and Results: The widely known Garfield (Impact) Factors (GF) are used or neglected, praised or criticized without any theoretical reasoning. The article shows that the GF can be regarded as a chance for citedness (number of citations per article). It is proved that the mean GF of journals of a field depends on the relative publication growth (i.e., ratio of the number of papers published recently to those issued earlier) multiplied by the mean number of references. It is shown that the normalized GF-s of journals of a set and the normalized specific impact contribution (SIC) indicators are identical measures. The SIC index relates the share of a journal in citations to that in publications within a set of journals.What's New? Scientometrics could hardly offer more appropriate elements than citations and papers for characterizing the impact of journals. The article concludes that the GF of a journal is a valid measure of its relative impact within a set of journals. GF may be regarded only formally as the mean citedness of the respective articles. Substantially, the GF of a journal measures its relative contribution to the total impact of journals devoted to the respective field. Accordingly, Garfield (Impact) Factors characterize the informational channels as entities.Limitations: Some of the findings refer only to the means of sets of journals. Assuming the calculations are correct, the article deals only with the theoretical background of GFs.
From JASIS&T v. 55 (6)
Jin, B., Li, L. & Rousseau, R. (2004). Long-term influences of interventions in the normal development of science: China and the Cultural Revolution, (544-550).
Study and Results: Intellectual and technological talents and skills are the driving force for the scientific and industrial development of a country. Major external events such as the Cultural Revolution in China or a prolonged period of war or social revolt, however, may have irreversible consequences on a country's scientific well being. Although the Cultural Revolution took place from 1966 to 1976, its aftermath can still be felt as shown in this study of the production and productivity of Chinese scientists as a function of their age. Based on the 1995-2000 data from the Chinese Science Citation database (CSCD), this article investigates the year-by-year age distribution of scientific and technological personnel publishing in China. It shows that the "Talent Fault" originating during the Cultural Revolution still exists. After 1979 China spent a lot of money and effort training young scientific and technological talent. The high peak of scientists in the age group between 30 and 40 shows that the younger generation has taken up the challenge and that they are the main force in the development of S&T in China. But we warn that a new gap resulting from the recent brain drain might be developing
What's New? The purpose of this work is to provide necessary information about the current situation and especially the existing problems of the S&T workforce in China. When the S&T personnel age structure becomes abnormal due to external events, it takes decades to return to normality, requiring several generations' efforts. Government policies can play a key role in reshaping such an abnormal situation by encouraging the young to enter the S&T system. The use of a local database (instead of ISI's data) was essential for this study, as the CSCD contains, besides the usual bibliographical information, data about the age of scientists. The article has a clear multidisciplinary focus, as it is partly sociological but based on typical information work (a scientific database) with implications for science policy.
From JASIS&T v. 55 (7)
Fan, W., Fox, E. A., Pathak, P., Wu, H. (2004). The effects of fitness functions on genetic programming-based ranking discovery for Web search, (628-636).
Study and Results: Previously, genetic programming (GP), an artificial intelligence technique, has been successfully applied to optimize a search engine's ranking function with the aim of improving Web search performance. However, in all prior work, only one fitness function was used in the experiments to guide the GP search. It's unclear how other fitness functions may impact the learning performance by GP. In this paper, we looked at the fitness function design from a utility theory perspective. Based on the utility theory, we designed several fitness functions that consider not only the relevance of documents, but also the order they appear in a search hit list – higher ranked relevant documents have higher utility values than those lower ranked ones. We applied these new fitness functions along with other existing ones to the search engine ranking function optimization task based on GP. The experimental results using 10GB TREC data indicate that the design of fitness functions is instrumental in performance improvement.
What's New? This paper makes a substantial contribution in providing a theoretical foundation based on utility theory for fitness function design, while prior research on the fitness function design mostly is designed commonly based on standard performance measures or human heuristics. The guidelines given in this paper will benefit other machine learning-based IR studies to better design their objective/fitness functions.
Scholer, F., Williams, H.E., & Turpin, A. (2004). Query association surrogates for Web search, (637-650).
Study and Results: In this study we propose and evaluate new techniques for effective Web searching through the creation of document surrogates. Our surrogates are based on query association, where queries that have a high statistical similarity to a document are stored as descriptors for that document. We evaluate our approach using a well-known TREC collection. Our results show that adding query associations to document content improves topic-finding Web searches by up to 7%.
What's New? We describe a novel approach to employ past user queries in the creation of document surrogates. Our experiments show that using query associations to augment document content can lead to a significant improvement in topic-finding Web searches. We also consider the effect of parameter settings and query log size on the effectiveness of our document surrogate approaches and demonstrate that performance is likely to improve further as more queries become available. Document replacements, where documents are indexed using only associated past user queries, are slightly less effective for retrieval, but can offer efficiency benefits due to the reduced size of the collection.
Limitations: Document surrogates created from query associations are effective for topic-finding searches, but less successful for site-finding searches. For the latter, a complementary technique, where surrogates are created from anchor text, can be employed.
Copyright © 2004, American Society for Information Science and Technology