A Learning Classifier System for Detection of Service-Level Agreement Violations in Business Process

Abstract

A service-Level Agreement (SLA) is a contract made between service providers and customers. This contract includes some constraints whose aim is to maintain a certain level of quality of service. A violation of such SLA constraints can be seen as a fault which means the failure of service guarantee. Since events in the system being analyzed are partially observable, this raises the most interesting case to address this problem. In this paper, a machine learning technique based on Learning Classifier System (LCS) to detect such violations is proposed. Thus, Anticipatory Classifier System (ACS) - a form of the LCS to deal with the case of partial observation is used. A typical example in a telecommunication company is used as a case study to generate the dataset utilized to train the model. The experimental results show a promising performance with an accuracy of 99%. In addition, the resulting model can give a timely decision whether a violation has occurred or not.