Research On AIS Based Service Matching Model

With regarded to the problem that the current system's self-learning ability usually appears weak during service discovery, a novel artificial immune system (AIS) based service match model is proposed, which employs the principles of cell's mutation, evaluation and the secondary response abilities, as well as simulation on the antibody-antigen identification mechanism.For the implementation of the system, firstly, self-study ability of the system is implemented, and the novel system can study by itself on the process of matching, and then, memorizes service matching results. With increase of matching times, the whole system efficiency is also improved. Secondly, matching results are divided into three parts (the highest match degree part, the hypo-high mach degree part and the other part), while the highest match degree part is regarded as request results, the hypo-high part is made full use of to evolve dynamically. Thirdly, the association ability (the service evolution ability) is also implemented. The hypo-high part is used to evolve dynamically to produce service requests, similar with current service request, and relevant services, thus, services produced by evolution doesn't need to experience primary matching and system will carry out secondary matching directly, so system efficiency is improved greatly.Theoretical analysis and simulations both show that the model can not only increase the recall ratio and match speed, but also realize the function that similar services can be obtained by known services. Furthermore, this service matching model is able to improve a system's self-learning, memory and dynamic evaluation capabilities.