An Effective Algorithm for Mining Weighted Association Rules in Telecommunication Networks

Abstract

The algorithms of weighted association rules mining and weights confirming were studied in alarm correlation analysis. A novel method named Neural Network based WFP-Tree (NNWFP) for mining association rules was proposed. NNWFP differs from the classical weighted association rules mining algorithm MINWAL (O). It is an efficient algorithm based on weighted frequent pattern tree, and the weights of the items are confirmed by the neural network. Experiments on a large alarm data set show that the approach is efficient and practical for finding frequent patterns in the alarm correlation analysis of telecommunication networks, and the performance of NNWFP is better than MINWAL (O).