Abstract:

A service automatic composition method based on discrete event calculus-Markov logic networks(DEC-MLN) is proposed aiming at the issues of service AI planning composition method such as a large number of services, incapability of handling dynamically created individuals during services’ execution and the uncertainty of business process and QoS in C⁴ISR systems. Firstly, the process chart of the C⁴ISR system on the naval fields is given and a probability extension of EC using MLN is presented. Then the basic semantic Web service composition processes and their IOPE are modeled based on the DECMLN. The uncertainty of the composition business process and QoS is described by the weight of MLN. The service automatic composition framework based on template is introduced and the abduction DEC planning method is given. The proposed method solves the confine of sequence composition process of the classic AI planning composition method, and the predicate number of the DEC-MLN is smaller, which speeds the composition. The DEC-MLN can handle the uncertainty and dynamically create individuals during services’ execution well, and it can also rule out the bound of the closed-world assumption. Finally, the experiment shows the feasibility and efficiency of the proposed method.