Efficient Mining of Recurrent Rules from a Sequence Database Using Multi-Core Processors

Software specification milling has been studied by many researchers. Among them, Non-Redundant Recurrent Rules Miner (NR3), proposed by Lo et al., mines recurrent rules. A recurrent rule can capture temporal patterns such as "Whenever a series of events occurs, another series of events must eventually occur", and it is particularly useful for program testing and verification to find bugs and ensure correctness of a software system. NR3 and its successor BOB are sequential algorithms for mining non-redundant recurrent rules, and mining recurrent rules still requires considerable computational costs. In this paper, we propose a new method, called pLF-NR3, to make NR3 more efficient; our approach is based on LF-NR3, an improved sequential algorithm of NR3, and it exploits parallelism existing in NR3. The proposed method is implemented on multi-core processors. We present some experimental results, which show the effectiveness of our proposed method.