A novel sequential pattern mining algorithm for the feature discovery of software fault

In order to obtain the useful sequential pattern knowledge from the historical sequence database, which reflects the characteristic behavior of software fault, a novel sequential pattern mining algorithm oriented feature discovery of software fault based on location matrix named SPM-LM is proposed. The pattern growth theory and the concept of location matrix are introduced into the new proposed algorithm. Firstly, the fault feature database is scanned and a location matrix for each event is constructed to record the frequent sequence information, which produces the frequent 1-sequence. Secondly, the sequence is extended through the dual pointer operation for the location matrix. And the frequent k-sequence for the prefix to frequent 1-sequence is generated. Finally, all of the generated frequent sequential patterns are saved into the corresponding layer of the tree structure. Therefore, the software fault sequences are matched in the tree structure to find the software failures and improve the software performance. The experimental results indicate that the algorithm improves the efficiency of pattern discovery significantly.