Qualitative preference-based service selection for multiple agents

Abstract

Service selection is one of the important problems in applications of multi-agent systems. A qualitative way is desirable for service selection according to agents' preferences on non-functional Quality of Service QoS attributes of services. However, it is challenging when the decision has to be made for multiple agents with preferences on attributes that may be incomplete. In this paper, first use a qualitative graphical representation tool called CP-nets to describe preference relations in a relatively compact, intuitive and structured manner. We then propose a preference reasoning algorithm to construct a derivation tree from a CP-net for each agent and then generate all service patterns for this agent. The Rank semantic is used together with the Lex semantic to provide the ordering of different service patterns. After that, we merge ranked service patterns for all agents and select a set of services that satisfy these agents the most. Finally, we also propose a semantic validation algorithm to show the satisfactory degree of the best service patterns according to other voting semantics. Experimental results indicate that this method can always obtain optimal outcomes which closely satisfy all agents, within acceptable execution time.