Product Configuration Using Case-Based Reasoning

摘要

This paper deals with product configuration, which attracts enterprises who carry on their business in electric-commerce. We propose a product configuration system that integrates constraint satisfaction problem (CSP) with case-based reasoning (CBR). First, when a user query is given, the CBR part retrieves similar cases from the case base in which past successful configurations are stored. Next, using the similar cases, CBR generates desirability criteria to evaluate the target products. Then, the CSP part solves the problem formalized with the query and the pans definition of the target product family, optimizing the products by the desirability criteria. Thus, since CBR acquires the user desirability of products, the products recommended by our system are appropriate to the user, even when the user specifies only a few conditions of the products. Simultaneously, these products are consistent with the product definition owing to the strict methodology of CSP. Moreover, our system provides a navigation function that is useful for users to achieve desired products interactively. Using the navigation function, the user can quickly find desirable products.