Application In Telecommunication Alarm Management, Data Mining Based On Association Rules

Alarm correlation system is an expertise system to analyze alarm data, which is an important part of telecommunication network management of alarm. However, it is difficult to acquire necessary knowledge to build an alarm correlation system of specific network because of its complexity. Data mining is a new means to find the knowledge of alarm information. In this paper, we research on data mining technology which is applied to alarm correlation system based on a telecommunication network management platform project of ZTE Corporation.Referring to some calculating models of alarm correlation system and algorithms of association rules mining, we analyze and build data mining system for mining alarm database of telecommunication network. There are three steps in mining process, including data collection and pretreatment, association rules discovery and association rules application. We also put forward the solution of related issues in implement process. The paper indicates that FP-Growth algorithm is more efficient than Apriori algorithm, with contrasting the two association rules algorithms. FP-Growth algorithm is intuitionist and easy for it only needs to scan database twice which decreases the I/O operation with high efficiency.The paper displays the key technology to implement data mining system on trait of telecommunication alarm data and the three steps of mining process. Data transfer is an important part of data collection and pretreatment, whose crucial part is to transfer continuous data to distribute data, so we use an inflexion partition method to implement it. We also extend FP-Growth algorithm to avoid large memory requirement by means of splitting alarm database to multi-databases with dealt data to realize date mining.It is feasible and significant that we use FP-Growth algorithm to implement data...