Daniel Griffin and Richard Alterman
The goal of this research is to develop technology for interactive case retrieval systems that improve their performance, after they have been deployed, by remembering previous questions. Depending on the origin of the case-base, one can expect a variable vocabulary and structure to the cases, leading to difficulties in determining exactly how the system should use the cases. By remembering questions, structure and uniformity can be built into the case-base, leading to improved system performance. In this paper, we describe a retrieval system that takes advantage of a question history. Experimental results are presented which confirm that this technique improves precision and recall of the system.