文章基本信息

标题：Automatically Inferring Structure Correlated Variable Set for Concurrent Atomicity Safety
本地全文：下载
作者：Long Pang ; Xiaohong Su ; Peijun Ma 等
期刊名称：International Journal of Software Engineering & Applications (IJSEA)
印刷版ISSN：0976-2221
电子版ISSN：0975-9018
出版年度：2013
卷号：4
期号：6
页码：51
出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：The atomicity of correlated variables is quite tedious and error prone for programmers to explicitly inferand specify in the multi-threaded program. Researchers have studied the automatic discovery of atomic setprogrammers intended, such as by frequent itemset mining and rules-based filter. However, due to the lackof inspection of inner structure, some implicit sematics independent variables intended by user aremistakenly classified to be correlated. In this paper, we present a novel simplification method for programdependency graph and the corresponding graph mining approach to detect the set of variables correlatedby logical structures in the source code. This approach formulates the automatic inference of the correlatedvariables as mining frequent subgraph of the simplified data and control flow dependency graph. Thepresented simplified graph representation of program dependency is not only robust for coding style’svarieties, but also essential to recognize the logical correlation. We implemented our method andcompared it with previous methods on the open source programs’ repositories. The experiment resultsshow that our method has less false positive rate than previous methods in the development initial stage. Itis concluded that our presented method can provide programmers in the development with the sufficientprecise correlated variable set for checking atomicity.
关键词：Correlated variables; frequent subgraph mining; program dependency graph