Dynamic update of experience for a Case-Based Reasoning system

Experience is one of the major factor of success of Case-Based Reasoning systems. This paper presents a learning algorithm that introduces reminding to update dynamically the experience of a system using a Reinforcement Learning model. Current research focuses on maintaining the experience growth by applying reduction techniques, but usually they do not consider adding new experience. For this reason, we propose a learning algorithm combined with two oblivion algorithms. All the algorithms are integrated into our model. Several experiments show the effectiveness of all the approaches in different domains from the UCI repository.