A case retrieval method of hybrid data based on information entropy
Jianfeng Hu, Jinhua Sun · 2016
According to the problem on non-uniformity of similarity measurement methods for different data types, and the difficulty of determining case attribute weights in the process of case retrieval method of hybrid data, the similarity measurement methods, including crisp symbol attribute, interval number attribute, crisp data attribute and fuzzy linguistic attribute, are respectively put forward. On the basis, a hybrid similarity measurement method is designed through the information entropy method to determine the case attribute weights, which is named IE-CBR. The example results show that the method can get the same calculation result which can also obtain by the method in classical literature. The conclusion is that the method can improve the efficiency of case based reasoning system by objective determination the weights and which also avoid the influence of subjective factors by artificially determining attribute weights in real data set testing.