A problem classification approach in business service management

IT is vital to business success or survival, IT provide not just technology but also a service. The objective of problem management is to minimize the impact of problems on the organization, which is an essential element for business service management. Supervised learning algorithms have been used for problem classification, but they rely on predefined classes and require each problem belongs to each category which sometimes is unreasonable. Moreover, with uncertainty, they may lead to poor quality. This paper proposes an uncertain fuzzy clustering problem classification approach based on partition. It does not rely on predefined classes and class-labeled training examples and allows that one problem can belongs to two or more categories by using fuzzy analysis and this approach use the expected sum of squared errors (E(SSE)) as its objective function instead of the sum of squared errors (SSE) to reduce uncertainty of data. This approach is unsupervised and automatic and is accurater than the others.