Feature-tree labeling for case base maintenance

Case Base Maintenance (CBM) algorithms update the content of the case base with the aim of improving the case-based reasoner performance. In this paper, we introduce a novel CBM method called Feature-Tree Labeling (FTL) with the focus on increasing the general accuracy of a Case-Based Reasoning (CBR) system. The proposed FTL algorithm is designed to detect and remove noisy cases from the case base, based on value distribution of individual features in the available data. The competence of the FTL method has been compared with well-known state-ofthe-art CBM algorithms. The tests have been done on 25 datasets selected from the UCI repository. The results show that FTL obtains higher accuracy than some of the state-of-the-art methods and CBR, with a statistically significant degreePeer Reviewe