A Potential Value Preferences Elicitation Approach Based on SC-VPM and KNN
Fang Zhou, Chao Ma, Jiaxing Qu, Xue Ling Song, Chi Zhang · 2020
Nowadays the more and more customers start to select and use the composite Web service on Internet, at the same time the services with the same functional properties but the different non-functional properties are increasingly emerging on Internet, which cause the information overload. Then the customer is not able to completely understand various composite Web services, and he/she is not able to define reasonable value preferences clearly on them. Therefore, this paper presents a potential value preference elicitation approach based on SC-VPM model and KNN algorithm, so as to support the third-party brokers to recommends top-satisfying services to customers according to the value preferences of the customers. In the approach, the inference rules based on the semantic relationships in SC-VPM model are used to preliminarily supplement the initial customer-value preference matrix firstly, so as to reduce the impact of the matrix sparsity on the following prediction. And then the KNN algorithm is used to identify the value preferences of K nearest neighbors customers, and the value preference vector of the target customer can be predicted and obtained. At last, a case is used to validate the proposed approach.