Intelligent Search for Correlated Alarms in Noisy Data

Title: Intelligent Search for Correlated Alarms in Noisy Data

Abstract: This research aimed to develop an efficient method for discovering correlated alarms in a database containing noise data. The study focused on defining two parameters, Win_freq and Win_add, to measure noise data and proposed the Robust_search algorithm to solve the problem. The algorithm was designed to discover more rules with a bigger size of Win_add at different sizes. The research also compared two interestingness measures, confidence and correlation, using experiments.

Main Research Question: How can we develop an efficient method for discovering correlated alarms in a database containing noise data?

Methodology: The study used data mining methods to analyze alarm correlation rules. It compared its approach with Mannila's WINE PI algorithm, which was applied to the TASA system. The research defined two parameters, Win_freq and Win_add, to measure noise data and proposed the Robust_search algorithm to search for correlated alarms in a database containing noise data.

Results: The experiments on alarm databases containing noise data showed that the Robust_search Algorithm could discover more rules with a bigger size of Win_add. The research also compared the interestingness measures of confidence and correlation, finding that both measures were effective in identifying important patterns.

Implications: The research has significant implications for the field of data mining and network management. It provides a robust method for discovering correlated alarms in a database containing noise data, which is crucial for improving the service and reliability of modern telecommunication networks. The study also contributes to the ongoing development of data mining techniques and their application in real-world scenarios.

Link to Article: https://arxiv.org/abs/0109042v2 Authors: arXiv ID: 0109042v2