An evaluation of case-based reasoning for fault diagnosis
Recalling past cases and their associated diagnoses to apply to a current fault situation is often an important step in the diagnostic process. In this thesis, a model for fault diagnosis in telephone networks has been developed, along with a prototype that implements the model using case-based reasoning. The prototype, called TETAA (TElephone Trouble Analyst's Assistant), aids in analysis and diagnosis of telecommunications faults by retrieving and presenting past fault cases and their diagnostic information. The diagnostic categorization model serves as an analytical tool to aid in fault reduction. The TETAA prototype was evaluated over a two month period at a New Brunswick Telephone Company (NBTel) repair center to determine its performance and potential in diagnosing and reducing network faults. The prototype demonstrated that case-based reasoning is applicable to fault diagnosis for a large telecommunications network and demonstrates potential for aiding in fault reduction.