A Hybrid Retrieval Strategy for Case-based Reasoning Using Soft Likelihood Functions
Yameng Wang,Liguo Fei,Yong Feng,Yanqing Wang,Luning Liu
DOI: https://doi.org/10.21203/rs.3.rs-561791/v1
2021-01-01
Abstract:Case-based reasoning (CBR) is the retrieval of one or more similar cases from an existing case base for the problem to be solved according to the characteristics of the new problem. The core idea of CBR is that similar cases have similar solutions, so whether the CBR system can play a powerful advantage depends on the quality of case retrieval strategy. At present, the commonly used case retrieval algorithm is based on the mean operator method, which is very hard, and a certain local similarity is low will affect the overall result. In order to calculate the global similarity of cases from a new and softer point of view, this paper introduces the soft likelihood functions into case retrieval, combines the soft likelihood functions with KNN, and proposes a hybrid retrieval strategy. The core of the retrieval strategy is to define the global similarity through SLFs, aggregate the local similarity and characteristic similarity together, and also take the attitude characteristics of decision makers into consideration. Through simulation experiments on real data sets, the accuracy rate is more than 81%, which verifies the effectiveness of the retrieval strategy.