The correction of software network model based on node copy

As the basic tool to study the complexity of software by using the theory of Complex Network, software network model is of great importance. Especially, class-collaborative network is superior in many fields and thus is widely used by researchers. However, as a type of static network reflecting the interactions between classes, the structure of class-collaborative network doesn't include enough information about dynamic binding, and thus misses some correlations between classes which are generated during the operation of software. To correct the class-collaborative network and let it can reflect the relations between classes generated by dynamic binding, we design a method called node-copy. To prove the correctness of our method, we use the node-copy method to generate software network models on four types of software different in structure obviously. For all network models above, by comparing the differences of spreading range and spreading paths before and after using node-copy method, we demonstrate that our method can effectively correlate the class-collaborative model and also reduces the complexity of the software network to ensure software security. Since our modified network model reflect the correlations between nodes (the structure of software) accurately and extremely, it can be used to study the propagation statistics of software and evaluate the importance of each node to the whole network.