Ph.D. Dissertation

Title

    A Concept Space Approach To Semantic Exchange

    Tobun Dorbin Ng
    Doctor of Philosophy in Management
    Management Information Systems
    The University of Arizona
    April 2000

Abstract

    This dissertation work investigates the use of information technologies that clarify semantic meaning to help users elaborate their information needs by providing library-specific knowledge to the information seeking process. The research involved two interdependent semantic technologies: concept space consultation and library-specific, domain-specific, automatically generated concept spaces.

    The concept space consultation phase used spreading activation algorithms - branch-and-bound and Hopfield net algorithms - to explore knowledge sources in specific domains. This research demonstrated the comparable effectiveness of exploration of a library database using a man-made classification scheme and thesaurus as opposed to an automatically generated concept space. The results showed that the use of spreading activation algorithms identified more relevant concepts than the use of the manual browsing method.

    The concept space technique automatically identifies and extracts concept from a library collection while at the same time computing the strength of associations between concepts. This research demonstrated that the concept space technique was able to create human-recognizable concepts and their associations. In addition, the technique could be scaled to generate very large library-specific concept spaces for a very large underlying library collection.

    Moreover, the interdependent use of both semantic technologies creates a semantic medium for users and library-specific knowledge sources to exchange content with context - context in user information need and that in corporeal knowledge.

Table of Contents

  • Brief Table of Contents
      Chapter 1 Introduction 14
      Chapter 2 Literature Review 17
      Chapter 3 Research Questions and Methodologies 56
      Chapter 4 Concept Space Consultation 69
      Chapter 5 Large-scale Concept Space Generation 122
      Chapter 6 Conclusions 162
      Appendix A. Benchmark Testing 173
      Appendix B. Sample Sessions 177
      Appendix C. Cancer Space: A Web-based Information Retrieval System 182
      Appendix D. Funding and Acknowledgements 189
      References 191

  • Detailed Table of Contents

Full Text

Citation

    Ng, Tobun Dorbin. A Concept Space Approach To Semantic Exchange. Ph.D. Dissertation, The University of Arizona. April, 2000.

Defense

Lessons Learned

    This dissertation has cumulated some research work performed over last decade (of the twentieth century) in the AI Lab. The research value and lessons learned were documented in various journal articles in addition to this dissertation.

    More importantly, it was a journey from a vague beginning of my career to a clear finishing of this dissertation. I was very fortunate to have unconditional support from my family and friends throughout this journey. They have elucidated me the meaning of life, love, and living.

      We are loved beyond our capacity to comprehend.
      Jewel Kilcher

Acknowledgement & Dedication

Remark on PDF

    All PDF files are converted from PostScript files using Adobe Acrobat Distiller 4.0. The conversion option used was Press Optimized. Hardcopy printing from these PDF files are excellent; the quality is as good as that from PostScript files. However, none of the conversion options give good screen-viewing quality on PDF files using Adobe Acrobat.

    Adobe Acrobat Reader is freely available.

    home       vita May 2000