Sensitivity analysis of web services in the problem of choosing the optimal system configuration with a service-oriented architecture

We study the problem of choosing proper measurement criteria for Web services' ranking. Existing approaches to selecting such criteria are reviewed and analyzed. The issue of Web services' sensitivity evaluation is studied in detail. The concept of sensitivity classes for web services is introduced. We suggest a routine based on machine learning techniques for isolation and automated class definition. We also present the results of a computer simulation that implements the suggested routine. In conclusion we outline promising lines of research.