Date of Award

Spring 2005

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

Committee Director

Kurt Maly

Committee Director

Mohammad Zubair

Committee Member

Frank C. Thames

Committee Member

Michael L. Nelson

Committee Member

Johan Bollen

Abstract

This dissertation studies the challenges and issues faced in federating heterogeneous digital libraries (DLs). The objective of this research is to demonstrate the feasibility of interoperability among non-cooperating DLs by presenting a lightweight, data driven approach, or Data Centered Interoperability (DCI). We build a Lightweight Federated Digital Library (LFDL) system to provide federated search service for existing digital libraries with no prior coordination.

We describe the motivation, architecture, design and implementation of the LFDL. We develop, deploy, and evaluate key services of the federation. The major difference to existing DL interoperability approaches is one where we do not insist on cooperation among DLs, that is, they do not have to change anything in their system or processes. The underlying approach is to have a dynamic federation where digital libraries can be added (removed) to the federation in real-time. This is made possible by describing the behavior of participating DLs in an XML-based language that the federation engine understands.

The major contributions of this work are: (1) This dissertation addresses the interoperability issues among non-cooperating DLs and presents a practical and efficient approach toward providing federated search service for those DLs. The DL itself remains autonomous and does not need to change its structure, data format, protocol and other internal features when it is added to the federation. (2) The implementation of the LFDL is based on a lightweight, dynamic, data-centered and rule-driven architecture. To add a DL to the federation, all that is needed is observing a DL's interaction with the user and storing the interaction specification in a human-readable and highly maintainable format. The federation engine provides the federated service based on the specification of a DL. A registration service allows dynamic DL registration, removal, or modification. No code needs to be rewritten or recompiled to add or change a DL. These notions are achieved by designing a new specification language in XML format and a powerful processing engine that enforces and implements the rules specified using the language. (3) In this thesis we explore an alternate approach where searches are distributed to participating DLs in real time. We have addressed the performance and reliability problems associated with other distributed search approaches. This is achieved by a locally maintained metadata repository extracted from DLs, as well as an efficient caching system based on the repository.

DOI

10.25777/15ae-a333

ISBN

978054236083

Share

COinS