Bulletin, April/May 2006

What's New?

Selected Abstracts from JASIST

Authors who choose to do so prepare and submit these summaries to the Editor of the Bulletin.

From JASIST v. 56 (10) 
(This issue was inadvertently skipped when it was published in the summer of 2005.)
Loureiro, O., & Siegelmann, H. (2005). Introducing an active cluster-based information retrieval paradigm, 1024-1030

Study and Results: When a client interacts with an expert, e.g., a doctor, following a basic introduction by the client, it falls upon the expert to ask questions that steer the process toward fulfilling the client’s needs. This is most efficient given that the expert has more knowledge about what information is available. On the other hand, most information retrieval systems leave to the user the task of coming up with queries. We propose an information retrieval framework that assumes the responsibility of steering the users to the information, thus increasing efficiency and user satisfaction.

What's New? Our active information retrieval system engages in a dialog with users to better estimate their information needs. It chooses optimally what relevance judgments to request from the user to minimize the number of questions. This dialog is personalized so that the user does not have to reply to too many or irrelevant questions. These features distinguish our system from many other relevance feedback methods for which the dialog may be far from optimal and overwhelming.

Limitations: Only multiple-choice questions are considered here. 

White, R.W., Jose, J.M., & Ruthven, I. (2005). Using top-ranking sentences to facilitate effective information access, 1113-1125.

Study and Results: In this article we describe content-driven information seeking, an approach to improve search effectiveness by encouraging a closer examination of retrieved search results. The approach uses query-relevant top-ranking sentences extracted from the top documents at retrieval time to shift the focus of interaction from document surrogates (such as titles and URLs) to document content, using this content to drive the information-seeking process. We evaluate the approach in three user studies, each applying these sentences in different ways; the findings showed that top-ranking sentences help searchers locate relevant information.

What's New? Content-driven information seeking is a new approach to interactive search that facilitates effective information access. Our user studies show that an approach that extracts, ranks and presents the content of documents (i.e., using top-ranking sentences) can be more effective than presenting ranked lists of documents, across different types of search task. The success of this approach has implications for the design of interfaces to search systems.

Limitations: Document collection: Although studies used the Internet the results are potentially generalizable to other domains.

From JASIST v. 56 (14)
Kwon, N., & Onwuegbuzie, A.J. (2005). Modeling the factors affecting individuals’ use of community networks: A theoretical explanation of community-based information and communication technology use, 1525-1543.

Study and Results: Despite a wide recognition of the service value, little is known about how people perceive and use a community ICT service. The present study investigated critical determinants of service use by constructing and assessing a theoretical model based on Ajzen’s (1985) theory of planned behavior. The model was tested using 417 community network users. Structural equation modeling identified the community network use model (CNUM), the final revised model, as the best fitting model. CNUM included seven predictors of use: three behavioral beliefs (i.e., learning, social interactions and community connection), normative beliefs-individuals, attitude, subjective norm and intention. The CNUM indicates that people used the service based on their beliefs in three functional utilities: learning, social interactions and community connection. These beliefs reflect the three most salient motivations that led people to use the service

What's New? This is the first study that investigated the constructs of TPB in the context of community ICT use. This approach is a substantial theoretical endeavor to explain use. Service practitioners can use the CNUM as an assessment tool for user needs and service effectiveness.

Limitations: Although the investigated community network represents a generic site, replications in other sites should be followed to consolidate or refute our findings.

From JASIST v. 57(1) 
Humphrey, S.M., Rogers, W.J., Kilicoglu, H., Demner-Fushman, D., & Rindflesch, T. C. (2006). Word sense disambiguation by selecting the best semantic type based on journal descriptor indexing: preliminary experiment, 96-113. Erratum in JASIST (2006), 57(5), 726.

Study and Results: This paper describes an experiment, including underlying methodology, in which journal descriptor indexing (JDI) was used for the disambiguation of words (like culture, transport and implantation) from the word sense disambiguation (WSD) test collection at the National Library of Medicine (NLM). JDI is ultimately based on statistical associations between words in a MEDLINE training set and a small set of journal descriptors (JDs) assigned by NLM to journals per se, which then form the basis for selecting the best meaning correlated to semantic types (STs) in NLM's Unified Medical Language System (UMLS). Overall average precision was 0.7873 compared to 0.2492 for a baseline method, and average precision for individual ambiguities was > 0.90 for 23 of them (51%), > 0.85 for 24 (53%), and > 0.65 for 35 (79%).

What's New? JDI-based WSD is an unsupervised method using ready-made resources at NLM. JDI uses these resources to automatically pre-label words in a training set with JDs and then with STs. Hand-tagging of a training set for word senses (as in supervised methods) is thus avoided.

Limitations: Ambiguities must be represented in the UMLS Metathesaurus. The WSD Test Collection contains only 50 ambiguities. The problem of not being able to assign the meaning “None of the Above” still needs to be resolved.

From JASIST v. 57(2)
Meso, P., Datta, P., & Mbarika, V. (2006). Moderating ICTs’ influences on socio-economic development with good governance: A study of the developing countries, 186-197.

Study and Results: In recent years, businesses in the United States and several other developed nations have faced very tough global competition from countries like China and India that can produce at much lower cost. At such lower production cost, China and India have captured a huge portion of the fast growing markets in developing nations calling for a mitigated euphoria for U.S. and European businesses that are interested in capturing the same markets. One disadvantage that managers in developed nations have is their relatively low understanding of the developing work and dynamics of navigating the complex socio-economic and political environments of these countries. With increasing globalization of the world economy, there has been a growing interest in the potential contributions of good governance to accelerating the rate of economic and social development in the developing countries and enhance their smooth integration into the emerging global economy. This study explores the link between information and communication technologies (ICTs), governance and social economic development in the developing countries.

What's New? Contributions of ICTs to social economic development are influenced by socio-political governance – leading to national development through more prudent and egalitarian application and use of the ICTs portfolio. The contingent role of governance on ICTs and national development paves the path toward recognizing the importance of socio-political moderators that are fundamental in establishing and running businesses in the developing nations.

Limitations: Data collection on information technology in developing countries is very constrained. In our own effort we found that no source provides complete data on both the range and reach of ICT variables, especially for computers and the Internet.