Enhanced Retrieval Processes for Knowledge Sharing

Peter Lyman and Hal R. Varian (2003), renowned for their work at Berkeley addressing information compilation and classification, estimated the amount of information on the public World Wide Web in the year 2003 to be 167 terabytes, or 17 times the size of the Library of Congress' printed collections. As this collection of information grows, it becomes more and more daunting to find the information needed from searches.

Compounding this dilemma is the fact that twenty percent of the content on the Web changes daily and another thirty percent changes every 30 days. As a result, current information retrieval technologies are inefficient mechanisms for searching, retrieving, and visualizing information. Among the problems are:

• Lack of context--search engines currently do not take into account the context and historical tendencies of the user when prioritizing the results of the information retrieval process.
• Lack of document quality measure--users have no way to determine the quality of a specific document without opening and examining it. They also generally cannot view or manipulate the standards by which commercial search engines rate documents.
• Lack of visualization and manipulation--most available search engines do not allow users to visualize or interact with the results of the information retrieval process such by filtering or re-sequencing the search results or redisplaying them in an alternative presentation (i.e., graphically).

Tim Wentling, Alan Craig, Steve Downey, Jennie File, and Andrew Wadsworth of the Knowledge and Learning Systems Group (KLSG) are working to address the above problems through the design, development, testing, and demonstration of new tools that will augment the existing Knowledge Center, a TRECC project combining knowledge management with e-Learning strategies.

To accomplish this, KLSG will develop and test intelligent software that learns about the user and his communities of practice in order to enhance the quality of information-based activities, such as processing search queries and conducting data mining activities. KLSG will also develop and test software tools that assess document quality for retrieval and that allow for manipulation of the assessment factors. Their final objective is the development of a user-controlled visual display of search display results that promotes user comprehension and learning.