Abstract: To solve the difficulties that the traditional web service selection based on quality of service (QoS) only considered the benefits (the mean) and not considered the risks (the dynamic changes), the service Skyline calculation method based on QoS descirbed by mean and standard deviation and four kinds of service Skyline query algorithms-block nested loop based on mean and standard deviation (BNL_MS), divide-and-conquer based on mean and standard deviation (D&C_MS), nearest neighbor based on mean and standard deviation (NN_MS) and branch and bound Skyline based on mean and standard deviation (BBS_MS) are proposed. This method can eliminate dominated service, can get the service set with excellent QoS and stable, and can effectively reduce the number of alternatives. Examples, experiments show that:1) the mean and standard deviation compared to the mean, interval number, and fuzzy numbers has better performance in the characterization of QoS; 2) Service Skyline query can effectively get the optimal service set; 3) BBS_CM compared to BNL_CM, D&C_CM and NN_CM has better performance.

Key words: mean and standard deviation, web service, benefits and risks, Skyline queries