Rule Discovery in Alarm Databases Rule Discovery in Alarm Databases

The papers in the series are intended for internal use and are distributed by the author. Copies may be ordered from the library of Department of Computer Science. Abstract Telecommunication networks produce large amounts of alarm information daily. This data contains potentially valuable knowledge about the network. We present a methodology for the analysis of large telecommunication networks alarm databases. The methods used aim at discovering useful knowledge about the network which can be employed in real time alarm handling software for ltering uninformative alarms, for correlating alarms to construct hypotheses of faults, or for fault prediction. The methodology is based on novel knowledge discovery methods for discovering patterns in alarm databases. We have implemented our methodology in the TASA (Telecommunication Network Alarm Sequence Analyzer) system which discovers patterns in alarm databases and provides tools for interactive identiication of the interesting patterns.