INFORMATION RETRIEVAL RESEARCH

PIs:

Margaret H. Dunham

Pete Fenner, Lightbus Technologies, Richardson, Texas

Students:

Badrinath Sampathkumar

Selecting correct query terms to search domain specific search engines is generally a difficult process, particularly if the user is not very familiar with the domain. This problem is evident in the case of patent search engines, where finding related patents is usually a difficult process. We have proposed a method of generating queries using a sample document as input, and identifying keywords that have a high probability of getting relevant results. The technique is to extract all possible keywords from the sample document and then filtering out the keywords that have low relevance based on a term weighting approach.

This research is supported by a gift from Lightbus Technologies in Richardson, Texas.