Visualizing and Exploring Document Spaces

  * Offered by the Information Access & Management group of the ICS
    department at UC Irvine
  * Customer contact: Dave Goggin  (daveg@ics.uci.edu)



David Goggin         www.ics.uci.edu/~daveg 
Yun Lan              www.ics.uci.edu/~ylan
  
BACKGROUND


  As technology and the Internet continue to bring ever-increasing amounts
  of information into the hands of the people, the tools that leverage
  human information-processing (particularly spatial visual perception)
  capabilities for understanding the structure of this information
  resource have not been sufficiently developed.

  Most Internet search engines present results textually, either in 
  categorical form, or as a long list of abstracted texts.  Neither of
  these formats show visually the space of documents, nor do they show
  the conformation of the document space by categories in a way that
  invites category exploration.  Likewise, conventional library search
  tools (e.g. ANTPAC, INSPEC) do not present a graphic interface to allow
  exploration of the collection, even when it has been indexed by human
  librarians into subject categories.

  Studies have shown that even when document collections are indexed by
  keywords, users typically do not use them in their exploration of 
  the document collections.  We hypothesize that this is mainly due to
  the lack of graphic support for such information foraging.

  Our system seeks to present a visual and explorable diagram of the 
  many-to-many relationship between keywords and documents, as well as
  a graphic display of that part of the category hierarchy relevent to
  the document collection, as well as supporting interactive document
  query and retrieval.   

  Begun in Summer '99, this document space visualization project has
  become an important research project for Dr. Wanda Pratt's group.  Our
  project is wildly interdisciplinary, drawing on insights from software
  engineering, theory of algorithms, cognitive science, and the fine arts.

BENEFITS TO STUDENT TEAM MEMBERS

  We expect we will learn a lot from one another!  In particular, you will
  have the chance to use some of the current generation of client-server
  tools for building web applications, as well as become skillful with
  Java and its associated libraries, tools, and graphics packages.  We'll
  probably also be learning some cognitive science to understand how
  information presented graphically is understood by users.  In addition,
  we will need to do some user-centered evaluation.  As an experienced
  software engineer, I will be able to teach you some of the theory and
  practice of object-oriented design and analysis, in particular, how to
  differentiate between a "good" and "bad" design.

PROJECT OUTLINE
    
  Mr. Goggin and Ms. Lan are currently putting together a prototype
  graphic depiction of the many-to-many relation between documents
  and keywords using multidimensional scaling (MDS).  We expect to
  demonstrate this prototype at the Jan. 28th ICS Research Symposium in
  addition to the present tree-based category graphic.

  There are essentially two parts of the project in which our ICS125
  team members will be involved:

  1. Finishing the graphical tools for displaying a document
  space visualization

    The present reused code for displaying the tree-based category graphic
    has very low capability because of a bug.  We need to either debug the
    code (by ourselves or in collaboration with the original developer) or
    redesign/reimplement it.
   
    Add zooming-in capability to the tree-based and MDS graphics.

    Do preliminary user testing and evaluation

  2. Extending the tool to allow graphical browsing and
  searching
  
    Designing efficient data structures and algorithms to support rapid
    updating of displayed graphics

    Designing and exploring alternative graphical representations

    Interfacing with other research projects (other graduate students' and
    Dr. Pratt's) to allow true real-time online searching through a web
    front end.

    User testing and evaluation of the system
 
  We expect to have our first team meetings soon.  But it is not
  likely that there will be much work to be done before Jan. 28th.

ICS125 WQ00 Project Opportunities
ICS125 WQ00
David F. Redmiles ­ Home Page
Department of Information and Computer Science
University of California, Irvine CA 92717-3425