Research On Multi-issue Symmetric Dependence Based On CP-nets

People tend to express their intentions through the preferences in everyday life, which assisted to make their own day-to-day decision-making. The seller also needs to know the preferences of the buyer to recommend satisfactory goods in e-commerce. So the expressive power of the language of preference statements affects the validity of finding the best assignment. In recent years, the research of describing the dependencies of issues effectively in multi-issue negotiation is a hotspot and difficulty. In this context, researchers have proposed many tools that can portray the dependencies of issues. Conditional Preference Networks (CP-nets) is a graphical tool which can describe the dependencies of issues. Its expressive power is very strong and it has natural, intuitive expression of the users’preference information. But it only describes, asymmetric dependence, and can not describe the continuous issues or some fuzzy, uncertain preference information, which is a limitation of the CP-nets.This paper describes the basic idea of dependency and the some various tools which can describe the dependencies of issues currently are analyzed and introduced in detail. This paper proposed a new graphical tool called Symmetric Dependence CP-nets(SDCP-nets) to describe dependencies of issues based on CP-nets because of various limitations of the existing tools. The tool makes up two limitations of CP-nets. First, SDCP-nets not only characterize the asymmetric dependence that is similar to CP-nets preference relations, but it also can characterize the symmetry dependence using utility matrix; Second, the tool use2-tuple linguistic model to describe continuous issues and uncertainty, fuzzy preference information to enrich the SDCP-nets’expressive power. In addition, the SDCP-nets structure, semantics and consistency are formalized and it can express the preference information more accurately, intuitively and naturally. Subsequently, this paper provides the algorithms of dominance choice and optimization choice. In SDCP-nets, the conditional preference relations make finding the best assignment more efficiently and the symmetric dependence relations ensure the accuracy of the algorithm results.Finally, the experimental verification has been conducted on the validity, accuracy, and the amount of inferring assignment preference relations. The experimental results show that using SDCP-nets to characterize the dependencies among issues can make find pareto optimal assignment effectively, and describe the users’preference statements with high accuracy, which make the algorithm results meet the expectations of the users more preferred.