A New Similarity Measure by Combining Formal Concept Analysis and Clustering for Case-Based Reasoning

This paper represents a new similarity measure by combining clustering, and Formal concept analysis (FCA). The novelties of this research are calculating the importance of features by clustering methods, and utilizing FCA to improve the accuracy of retrieving similar cases. Also, using the FCA helps us, manage the case-base structure. Finally, after performing several experiments on the UCI datasets with cross validation, our new similarity measure improve the accuracy of classification CBR significantly when compare to the other six measures proposed before.