Software Fault Localization Model Based On Complex Network

With the advent of the Internet period, the software has been applied to differentaspects of people’s lives, a variety of PC and mobile client software is changing the waypeople communicate. The software is moving the direction of scale expansion anddiversification of relations in software design process, in which different target objects arebecoming more closely. The complex software systems are developed, and they are nolonger simple mapping relations so the software security issues are taking more attentioninto account. For that, reducing the software fault is the key to ensure the software thequality in the software development. From the perspective of the network, the papercreates a software dynamic graph-based analysis modeling and designs a complexnetwork-based key node selection method which improves the result of software faultlocalization.First, considering the software dynamic graph model, we propose a mining featurenodes method related to software fault. First, we create software dynamic execution graphbased on software test cases, then divide graph collection into pass and fail ones from theresults of test cases. Finally, we filter out the pass and fail graph in use of the comparativeanalysis. Through the differences in two ones, the feature nodes of software fault areindentified and help software developers to locate the fault region.Secondly, from the viewpoint of complex networks, we propose a suspicious nodestrategy and design metrics of software critical nodes indicating the importance of a node.Analyzing the relation of node suspicious weight and network execution path, our strategycombines each test case execution graph and analyzes them. This strategy calculates eachnode suspicious weight. Meanwhile, the key node measure of complex network is appliedto sort the feature nodes, indicating the fault severity of suspicious nodes.Finally, the paper makes related experimental verification, experimental results showthat the proposed method can improve the success rate of software fault localization andthe sorting effectiveness, help software developers to reduce the code inspection costs, andenhance software security protection.