JASIST IndexJASIST Table of Contents

Journal of the Association for Information Science and technology

 EDITORIAL

 

In This Issue
   
Bert R. Boyce
 

877

 

 RESEARCH

 

The Effect of Extrinsic Motivation on User Behavior in a Collaborative Information Finding System
    Bracha Shapira, Paul B. Kantor, and Benjamin Melamed
    Published online 13 July 2001

In collaborative information systems similarity is calculated between new queries and existing ones for which relevance feedback data has been collected. This implies that there will be significant relevance feedback with users contributing their evaluations of material received.

Shapira, Kantor, and Melamed examine motivation for such contribution in the AntWorld collaborative and conclude that some extrinsic motivation to contribute is required. Students were given course assignments requiring system search and asked for online feedback during the sessions. In the test without external incentives 0.8 judgments were provided per question. In a later test the group with the highest number of judgments was promised a pizza party. This test produced 2.3 judgments per query and the students visited more pages overall.
 

879

 


 

 

 

 

 

 

 

 

Using Clustering Techniques to Detect Usage Patterns in a WebBased Information System
   
HuiMin Chen and Michael D. Cooper
    Published online 19 July 2001

Chen and Cooper examine the transaction logs of the web-based  MELVYL online catalog after removing tourist type interactions, in order to classify usage sessions into like groupings, using principle components analysis for data reduction and cluster analysis to create user groups. Six distinct usage groups were identified: knowledgeable and sophisticated use, unsophisticated use, highly interactive use with good search performance, known item searching, help-intensive searching, and unsuccessful use.

Vectors of forty seven variables concerning times, character of search statements, displays and relevance information based on display activity, were extracted from two 1998 Spring semester samples, one of 126,925 sessions, the other of 130,902. A principle components analysis yielded 16 components with eigenvalues greater than one that together explain 76% of the variance in the original data and these vectors were used in a cluster analysis. SAS FASTCLUS, a nonhierarchical method was used to form 100 classes, and after the removal of 16 out lying clusters, the remaining 84 clusters were passed through SAS CLUSTER using Ward's algorithm, a hierarchical method. SAS computes the pseudo statistic t2 to assist in choosing the optimal number of clusters. A large increase in t2 indicates the optimum number of clusters and resulted in 6 clusters with both samples. Sample two components were then assigned to the new six centroids from sample one, and evidence is presented that this clustering is in agreement with the other method. The clusters were then perused qualitatively to determine their character.

The Arrowsmith software takes two sets as input and produces a list of terms common to both and when run on the pathogenetic set and each of the other three will produce a list of common subject headings for each. From this list the virus names can be culled and listed indicating which and how many set pairs produced them. Such a list provides the raw material for human evaluation of candidate viruses. To test the technique such a list was compared to a previously produced list of possible BW viruses and was found to contain most of these, with a significant relationship inferring the new virus names to be similar to the old. It would appear that a virus can be classified as a potential weapon based upon literature structures associated with it.
 

888

 

 

 

 

 

 

 

 

 

 

 

 


 

 

 

 

 

 

Perspectives on Image Access:  Bridging Multiple Needs and Multiple Perspectives
    Introduction and Overview
    Corinne Jorgensen
    Published online 10 August 2001

For the last few centuries, words have been the privileged form of communication and the preferred means of education. A shift has taken place, however, within the last several decades, and images have been reasserting their primacy as immediate and influential messengers. Fortunately, with the exponential growth in digitized collections of images, indexing of images has become an important research area over the last ten years.

There have been two general approaches to image indexing automated and manual. Automated (often known as "content-based") indexing produces primarily numeric representations of image content and permits retrieval by low-level image features, such as color, texture, and shape. Manual (or "human-based") indexing assigns textual indexing terms (free-text or controlled vocabulary) to visual materials. Content-based research generally focuses on algorithms, while indexing research looks at the types of attributes to be indexed, creation of indexing vocabularies, and the needs of particular user groups.

This Perspectives issue is motivated by the need to link these research communities and heighten awareness of projects utilizing an expanded theoretical base and new techniques for representation. This collection of articles brings to the fore issues of major concern to both communities.

Although many of the articles in this issue address issues of system design, each of the authors is concerned with multiple areas within the research agenda. Indeed, it is this intertwining of the image, system, and user which makes this research particularly challenging, but also, in the long run, has the potential to produce truly usable and useful image retrieval systems.

Within this issue, the articles are in three groups, focusing on user, representation, and system issues (please note, however, each of these articles has significant components of the several areas addressed, and many common themes are interwoven throughout them). Roberts sets the stage for the articles that follow by addressing issues of user needs. Greenberg looks more specifically at the users' tasks and the effects of domain in her functional analysis of metadata. The next group of authors focuses on the image and methods for structuring its representation (Gordon, Tam & Leung, Jörgensen et al.), and manipulation and extraction of existing representations to create new ones or additional access points (Goodrum et al.). System-focused articles describe system architecture for multimedia databases (de Vries) and evaluation of retrieval systems (Smith, Müller).
.

    906

 

 

 

 

 

 

 

 

 


 

 


 

 

 

 

 

 

 

 

 

 

A Picture is Worth a Thousand WordsArt Indexing in Electronic Databases
   
Helene E. Roberts
    Published online 10 August 2001

Roberts discusses the needs of art historians but her points are widely applicable to image searching in many domains. She presents a strong argument for access to both the visual content and meaning of an image, as well as to the concepts and contextual information surrounding the image. A fundamental point is that an understanding of the structure of the archive (system) itself and the context within which it was created reveals the limitations imposed upon a collection (and access to its contents) by the principles underlying its formulation.
   

    911

 

 

 

 

 

 

 

A Quantitative Categorical Analysis of Metadata Elements in ImageApplicable Metadata Schemas
    Jane Greenberg
    Published online 10 August 2001

Greenberg's functional analysis of existing metadata demonstrated major differences in the schemas because of variability in domain (the professional impetus for image creation) and user (the task for which the schema is applied). Her methodology provides us with a foundation for structural system design based upon an understanding of user tasks and provides a new basis for developing interoperability among systems.
 

917

 

 

 

 

 

 

Browsing Image Collections with Representations of CommonSense Activities
    Andrew S. Gordon
    Published online 10 August 2001

Gordon's article addresses themes of access and evaluation within user-focused design of a browsing system, which should permit variation of access methods depending on the task of the user. Retrieval systems are frequently designed to privilege specific item searching at the expense of exploration, and current methods of system evaluation generally do not address the browsing capabilities of a system. Gordon used a conceptual approach to modeling image content, as do Tam and Leung and Jörgensen et al.
 

925

 

 

 

 

 

 

Structured NaturalLanguage Descriptions for Semantic Content Retrieval of Visual Materials
    A. M Tam and C. H. C. Leung
    Published online 10 August 2001

Tam and Leung are also concerned with representation of a full range of image attributes within a single searchable structure. They describe a conceptual basis for solving the problem of relating low-level features to higher-level semantic interpretations. Building upon Roberts' ideas of the importance of other textual documentation as a source for further information, they propose linking to external resources such as thesauri and ontologies for additional contextual knowledge.
 

930

 

 


 

 

 

 

A Conceptual Framework and Empirical Research for Classifying Visual Descriptors
    Corinne Jorgensen, Alejandro Jaimes, Ana B. Benitez, and ShihFu Chang
    Published online 10 August 2001

Jörgensen et al. present a structure for representing a full range of image attributes, incorporating multiple syntactic and semantic levels. They share with the other authors the purpose of providing tools for the disambiguation of keyword searching and a common structure which can accommodate both textual and content-based descriptors.
 

938

 


 

 

 

 

An Open Source Agenda for Research Linking Text and Image Content Features
    Abby A. Goodrum, Mark E. Rorvig, KiTai Jeong, and Chitturi Suresh
    Published online 10 August 2001

Goodrum et al. provide a research agenda which bridges the communities involved in image research by utilizing automatic methods to accomplish some of the traditional goals of classification. They also propose making open source software widely and freely available to the broad image retrieval research community which could further the research agenda proposed here and also contribute to other benchmarking efforts as described by other authors in this issue.
 

948

 

 

 

 

 

 

Content Independence in Multimedia Databases
    Arjen P. de Vries
    Published online 10 August 2001

de Vries addresses system design issues, specifically the functionalities needed for management of multimedia digital library collections. de Vries shares Gordon's concerns for the restrictions imposed by predefined access patterns and other authors' concerns with providing a unified structure which can accommodate multiple levels of description. Related to the suggestions of Goodrum et al., a system associating textual terms to content clusters has been designed.
 

954

 

 

 

 

 

 

Evaluating Image Browsers Using Structured Annotation
    Wolfgang Muller, Stephane MarchandMailet, Henning Muller, David McG. Squire, and
    Thierry Pun
    Published online 10 August 2001

Müller et al. and Smith address system evaluation issues. Müller et al. share Gordon's concerns for evaluation methods for browsing systems, but within the context of content-based retrieval systems rather than textually indexed collections. They seek an automated way of simulating user needs to a system to reduce the overhead involved in such evaluation, which often involves conducting large scale user studies. Additionally, performance measures must be defined within the context of user behaviors. The proposed system could be used as a tool to explore the relationships between low-level features and higher-level semantic descriptions.
 

961

 

 

 


 

 

 

 

  Quantitative Assessment of Image Retrieval Effectiveness
    John R. Smith
    Published online 10 August 2001

Smith's article presents a typology of image dimensions based on structure, image content (features), and meaning (semantics) and an overview of concepts associated with these and with content-based retrieval systems. He also describes a major impediment to further progress in image retrieval systems research the lack of methods of quantitative system evaluation which will produce comparable results across systems. As a solution, he proposes a TREC-like standardized test-bed for benchmarking image retrieval systems and describes the components of such a system.
 

969

 


 

 

 

 

BOOK REVIEWS

 

Creating WebAccessible DatabasesCase Studies for Libraries, Museums, and Other Nonprofits, by Julie M. Still
    Lisa A. Ennis
    Published online 6 July 2001
 

980

 

 

 

Community InformaticsEnabling Communities with Information and Communication Technologies, edited by Michael Gurstein
    Tatyana Dumova
    Published online 18 July 2001

981


 

 

Academic Libraries as Hightech GatewaysA Guide to Design & Space Decisions, 2nd Edition, by Richard J. Bazillion & Connie L. Braun
    Ina Fourie
    Published online 18 July 2001
 

983

 

 

 

Editorial Peer Review:  Its Strengths and Weaknesses, by Ann C. Weller
    Dale A. Stirling
    Published online 6 July 2001
 

984

 

 

CALL FOR PAPERS

986


ASIST Home Page

Association for Information Science and Technology
8555 16th Street, Suite 850, Silver Spring, Maryland 20910, USA
Tel. 301-495-0900, Fax: 301-495-0810 | E-mail:
asis@asis.org

Copyright © 2001, Association for Information Science and Technology