An efficient approach to mine high utility path patterns from software executing traces

作者：Ren Jiadong; Zhou Yang^*; Wang Jiandi; Xu Cui; Yue Liwen

来源：International Journal of Innovative Computing Information and Control, 2016, 12(5): 1453-1465.

• High utility
• Improved structures
• Novel structures
• Pattern mining
• Program comprehension
• Sequential patterns
• Software behavior
• Synthetic datasets

Studies have shown that software behavior mining is an important task in program comprehension. Certain characteristics of software traces are often hidden and can be analyzed from some desirable patterns. Therefore, finding these interesting high utility software executing patterns is meaningful. In this paper, an efficient technique called FHUPPM (faster high utility path patterns mining) is proposed to mine high utility path patterns from software executing traces. Firstly, since the initial software executing traces are neither complete nor intuitive, we need a preprocessing on them to get the unique software sequential patterns. Secondly, FHUPPM uses a novel structure, called pattern-index utility list shorted as PIUL, to store both patterns index and utility information for pruning the search space and extending adjacent path patterns. What is more, an improved structure named UCMS based on the analysis of item co-occurrences is put forward. Then an efficient pruning strategy which relies on the UCMS can be applied to eliminating the low-utility patterns directly. Finally, we compared FHUPPM with some previous algorithms on both real and synthetic datasets. The experimental results show that our algorithm outperforms them especially on time.

• 出版日期2016