The Research Of Concurrent Program Bug Detection Technology Based On Sequential Pattern Mining

For the main bugs existing in the concurrent programs, the present study includes a single type of bug detection technology and general variety bug detection technology. But these technologies have their respective shortcomings such as producing the combination explosion problem or higher false alarm rate. So the technologies don’t meet the requirements of concurrent programs in bug detection well. And now the study of concurrent program bug detection mostly stays in a single type of bug detection technology, and the research for general variety bug detection technology is still insufficient. Sequential pattern mining method is applied to concurrent program in bug detection, so it can detect many kinds of bugs, and it can solve the combination explosion problem effectively so that it improves the efficiency of bug detection.In order to improve the safety and reliability of concurrent programs, to facilitate program debugging, the paper proposes a concurrent program bug detection technology which based on Sequential Pattern Mining(SPM), the contributions are:First of all, the paper introduces the overall framework of concurrent program bug detection technology based on Sequential Pattern Mining, analyzes the relations between obtaining the concurrent program execution path, bug detection based on Sequential Pattern Mining and the experiment design and sums up the main purpose and design idea of each part.Secondly, the paper studies the specific methods of obtaining the concurrent program full execution path set. The paper uses the Improved DPOR algorithm to explore all thread staggered implementation and uses AspectT to record thread and events information to reflect the program execution path at execution time, then tests program execution results. According to the execution results, the program execution sequence set is divided into successful execution sequence set and failure execution sequence set.Then, the paper studies how to use the Sequential Pattern Mining to get bug patterns. After obtaining the program execution path, the paper studies how to use data mining for program execution path set to get bug patterns. And the paper has made the improvement of sequential pattern mining algorithm and proposes ISPAM algorithm to solve the problem of requiring the user to set special support and low efficiency of the algorithm.Finally, the paper carries out the experiment on artificial data sets and real data sets based on the above frameworks and methods.On the basis of the study, the paper constructs the corresponding experimental environment, carries out the experiment and analyze the experimental results of the proposed method of the paper and the traditional method. And the experiments show that the proposed methods of the paper are effective.