摘要

To implement uncertain QoS-aware Web service selection under the circumstance of the deficiency of user's precise utility function and the unknown of QoS probability distribution forms, an admissible composition plans based on stochastic dominance theory was proposed to discover the function class which were compatible with the user utility. According to the expected utility theory, the admissible composition plans could not be dominated by any others. By eliminating the services or intermediate composition plans which were included in the admissible set, the search space in finding the admissible set was significantly reduced. Moreover, a heap-based composition plan extracting and expanding process was developed based on the properties of stochastic dominance, so that the numbers of stochastic dominance tests were reduced. The effectiveness and efficiency of the proposed method were verified by theoretical analysis and comprehensive experimental research based on real datasets of Web service QoS.

全文