Preference-Based CBR: A Search-Based Problem Solving Framework

Abstract

Abstract. Preference-based CBR is conceived as a case-based reasoning methodology in which problem solving experience is mainly represented in the form of contextualized preferences, namely preferences for candidate solutions in the context of a target problem to be solved. This paper is a continuation of recent work on a formalization of preferencebased CBR that was focused on an essential part of the methodology: a method to predict a most plausible candidate solution given a set of preferences on other solutions, deemed relevant for the problem at hand. Here, we go one step further by embedding this method in a more general search-based problem solving framework. In this framework, case-based problem solving is formalized as a search process, in which a solution space is traversed through the application of adaptation operators, and the choice of these operators is guided by case-based preferences. The effectiveness of this approach is illustrated in two case studies, one from the field of bioinformatics and the other one related to the computer cooking domain.

Case-Intelligence Recommendation on Massive Contents Processing through Dynamic Computing

“…Normally, recommendation system uses low level analogy reasoning, which is an important cognitive model of human sense [7]. The "low level" means simple analogy, which is not inferring in different domain data.…”