Causal Query Elaboration in Conversational Case-Based Reasoning

A key research focus for conversational case-based reasoning (CCBR) is incremental query elaboration, which is the process of maximizing the extraction of relevant problem state information throughout the querying process. Several companies and researchers have addressed this problem (e.g., by dynamically applying domain-specific plans (Carrick et al., 1999)). Recently, Gupta (2001) demonstrated how CCBR can be significantly enhanced through the judicious use of (1) taxonomies to represent domain information and (2) a control algorithm for focusing case retrieval. However, in that original conception, the individual taxonomies were isolated from each other, and from other information sources that could support query elaboration. This prevents information from being propagated to these taxonomies, and could inflate the length of the user’s problem-solving session. In this paper, we outline and exemplify a causal query elaboration method for inter-taxonomy communication and highlight its potential benefits, which include shorter (and potentially more accurate) conversations, support for causal inferencing, and more concise case representations.