We look at the effect of hitlist size in a Question-Answering system. When the goal is to find a single answer to a fact-seeking question, it becomes readily apparent that looking at too many documents can be a source of more noise than useful information; looking at too few documents can miss the desired answer. We develop a probabilistic model of the Answer-Selection component that extracts candidate answers from passages returned by the search engine. We show with this model: how we can predict performance as a function of hitlist size after making observations of the system’s operation using a single such size; that an optimum hitlist size exists; how the performance depends on a parameter we call the discrimination ratio, and the benefit to be derived from improving this ratio by more training. We show that the performance curves generated by our model agree very well with empirical results.
By: John M. Prager
Published in: RC22314 in 2002
LIMITED DISTRIBUTION NOTICE:
This Research Report is available. This report has been submitted for publication outside of IBM and will probably be copyrighted if accepted for publication. It has been issued as a Research Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IBM prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g., payment of royalties). I have read and understand this notice and am a member of the scientific community outside or inside of IBM seeking a single copy only.
Questions about this service can be mailed to reports@us.ibm.com .