A Framework for Mining and Applying Alarm Rules Based on LLM

Driven by 5G technology, network performance and management requirements are becoming increasingly sophisticated. However, massive and complex alarm information is also coming, posing severe challenges to network operation and maintenance. Accurately identifying alarm correlations and locating alarm root causes has become a core problem for operation and maintenance. In response to this challenge, this paper proposes a framework for mining and applying alarm rules based on large language models, which applies the text understanding capabilities of large language models to alarm processing and can mine uncommon alarm rules. Experimental results show that this framework significantly improves the accuracy and coverage of alarm rules, and improves the response speed and efficiency of network operation and maintenance.