Edie Rasmussen

User's relevance criteria in image retrieval in American history

A large number of digital images are available and accessible due to recent advances in technology. Since image retrieval systems are designed to meet user information needs, it seems apparent that image retrieval system design and implementation should take into account user-based aspects such as information use patterns and relevance judgments. However, little is known about what criteria users employ when making relevance judgments and which textual representations of the image help them make relevance judgments in their situational context.Thus, this study attempted to investigate the criteria which image users apply when making judgments about the relevance of an image. This research was built on prior work by Barry, Schamber and others which examined relevance criteria for textual and non-textual documents, exploring the extent to which these criteria apply to visual documents and the extent to which new and different criteria apply. Data were collected from unstructured interviews and questionnaires. Quantitative statistical methods were employed to analyze the importance of relevance criteria to see how much each criterion affected the users judgments. The study involved 38 faculty and graduate students of American history in 1999 in a local setting, using the Library of Congress American memory photo archives.The study found that the users perception of topicality was still the most important factor across the information-seeking stages. However, the users decided on retrieved items according to a variety of criteria other than topicality. Image quality and clarity was important. Users also searched for relevant images on the basis of title, date, subject descriptors, and notes provided. The conclusions of this study will be useful in image database design to assist users in conducting image searches. This study can be helpful to future relevance studies in information system design and evaluation.

Evaluating interactive systems in TREC

The TREC (Text REtrieval Conference) experiments were designed to allow large‐scale laboratory testing of information retrieval techniques. As the experiments have progressed, groups within TREC have become increasingly interested in finding ways to allow user interaction without invalidating the experimental design. The development of an “interactive tract” within TREC to accommodate user interaction has required some modifications in the way the retrieval tasks is designed. In particular there is a need to simulate a realistic interactive searching task within a laboratory environment. Through successive interactive studies in TREC, the Okapi team at City University London has identified methodological issues relevant to this process. A diagnostic experiment was conducted as a follow‐up to TREC searches which attempted to isolate the human and automatic contributions to query formulation and retrieval performance.

Developing a new similarity measure from two different perspectives

In this paper two distinct similarity measures in a document vector space, the distance-based and angle-based similarity measures, are compared, and a newly developed similarity measure based upon both the distance and angle strengths of two compared objects is presented. The concept of the iso-extent contour, which facilitates the understanding of the nature of the newly developed similarity measure, is introduced. The three different similarity measures are compared and the properties of the newly developed similarity measure are addressed.

A digital library for education: the PEN‐DOR project

Recent initiatives in digital library research have suggested new models for the creation and organisation of digital information and its dissemination to virtual communities. PEN‐DOR (the Pennsylvania Education Network Digital Object Repository) is a digital library designed to provide access to the collective experience of teachers, students and administrators in public schools in building lesson plans and using curriculum materials. Using the WWW as a platform, PEN‐DOR incorporates current research in digital libraries to provide K‐12 educators with access to multimedia resources and tools to create new lesson plans and presentations, and to modify existing ones. Design problems addressed by the project include the design of a distributed, object‐oriented database architecture, the description and cataloguing of multimedia objects, and issues related to usability and training for a geographically scattered user community. Two critical aspects of the organisation of this digital library are the developm...

Introduction: parallel processing and information retrieval

Abstract Recent advances in parallel processing have made powerful computing resources available for information retrieval. This introduction, an overview of trends and work in parallel information retrieval, covers three areas: the development of parallel algorithms for information retrieval, the design of parallel architectures to handle information retrieval tasks, and the distribution of databases over multiple processors. The articles in this issue are introduced in the context of ongoing work.

Metadata for a Digital Library of Educational Resources

SUMMARY PEN-DOR (the Pennsylvania Education Digital Object Repository) is a digital library providing access to atomic Web-based objects for lesson plan construction, a set of fully constructed lesson plans, and curriculum standards for the state of Pennsylvania. PEN-DOR supports lesson plan construction and enhancement activities. Through a community-based memory documentation process, PEN-DOR plans to provide access to the collective experience of teachers, students, and public school administrators working with the repositorys resources. The diverse activities supported by PEN-DOR present a series of challenges in organizing and accessing Web-based objects, lesson plans, and other PEN-DOR resources for use. This article focuses on the development and implementation of the PEN-DOR metadata scheme, and discusses a number of metadata-related challenges that have emerged as a result of the project.

An experimental study on the iso-content-based angle similarity measure

An experimental study was conducted to examine and evaluate retrieval performance of the iso-content-based angle similarity measure presented in an earlier paper. Retrieval performance of the iso-content-based angle similarity measure within the angle, distance, conjunction, disjunction and ellipse retrieval models is compared with retrieval performance of the distance similarity measure and the angle similarity measure. The experimental results show the iso-content-based angle similarity measure achieves satisfactory performance.

Visualizing knowledge domains. Sponsored by SIG CR, SIG VIS

Information visualization can be a powerful tool for simplifying access to complex material. This panel will explore the use of visualization techniques to organize and display the structure of knowledge in subject domains, and the extent to which it is successful in clarifying the scope of individual fields and the relationships between concepts within fields and with related fields. New work in topic maps and other visualization techniques for scientific disciplines will be presented.