Troubleshooting network faults using past experience

Abstract

So far, trouble ticket systems have been used to store information concerned with network faults. An interesting approach is the development of an expert system that takes into account the knowledge accumulated by these systems in order to propose solutions for a new problem. This work presents DUMBO, a system that applies the case-based reasoning paradigm to a trouble ticket system in order to aid the diagnosis of computer network problems. One important aspect of DUMBO is that it performs continuous learning, i.e., the knowledge base is increased by the incorporation of new solved cases in a way such that other similar cases can be proposed as solutions to a current new problem. Another important aspect is the use of production rules in some specific tasks of the reasoning processes. This work also describes typical domain problems, the system architecture, the experiments and practical results reached with the developed prototype and its current state.