Case indexing in Case-Based Reasoning by applying Situation Operator Model as knowledge representation model

Case-Based Reasoning (CBR) is an effective technique for solving cognitive problems. It maintains old experiences which are encountered in various problems and presented as cases in a case-base. Consequently, indexing the cases in memory is an important issue in enhancing retrieval and learning performance in a cognitive system. This paper focuses on this issue and proposes an approach by integrating fuzziness for case indexing in CBR. Generalization and fuzzification of similar cases are the main properties of this approach in improving the performance of retrieval as well as reducing the size of case-base. Also, the connection between signal-based measurement level and problem-oriented behavioral level of the system which is a missing level in actual approaches is realized from a principle point of view in this study. In addition, Situation Operator Model (SOM) as a knowledge representation model is applied to CBR to model the events, related actions, and their effects in term of cases. As an application experiment, COLIBRI is utilized as a CBR reference platform. A data set generated using a driving simulator is utilized to learn two different classes of cases (start passing, end passing) generated from maneuvers of drivers. Finally, the performance of retrieval process and similarity error on fuzzy and conventional indexing approaches are measured to reveal the effectiveness of the presented approach.