Experiments on information Retrieval using Case-Based Reasoning

Abstract

This paper presents a demonstration of a Theory for Episodic Learning applied to Information Retrieval. The theory is compatible with Case-Based Reasoning and is used to model how contextual information about users, their aims and search-sessions experience may be represented in case bases, and automatically tuned up. The proposed model shows how query processing may be extended by deriving information from the case bases, taking advantage of user profiles and a history of queries derived from search-sessions. The paper includes a demonstration of the program IRBOC, presenting a fully worked example. © Springer-Verlag Berlin Heidelberg 2000.