The Analysis And Application Of Ripple Degree In Software Networks

The science of complex system has a view that the topological structure of system will affect its function, performance, reliability and other system indexes. Software product as a kind of artificial intelligence system is also appearing the same. Software engineers not only have to pay attention to system functions, but also need to understand the structure of software, so that more software with good constructivity and evolvability can be developed. Recent years, academia used complex network widely to explore the internal structure of software, and some fine results have been obtained. However, those software complexity metrics have been provided not perfect yet and still exist deficiencies. This thesis used complex network to research the complexity of software structure. This is of importance to the risk assessment, quality control and maintainability of software.Reversal and forward ripple degrees are used in this thesis to measure the affecting scope and affected scope of software modules. Through the research, find that reversal and forward global ripple degree exaggerates module’s affected scope and being affected level to some extent. So, reversal and forward local ripple degree are introduced in this thesis.In the comprehensive explore of reversal and forward local ripple degree’s distribution of every node, discover that three kinds of node(with large reversal and small forward local ripple degree, or with small reversal and large forward local ripple degree, or large reversal and large forward local ripple degree) exist in software network. In order to measure out the three kinds of nodes, three measuring methods are proposed in this thesis. These three formulas are used in the following part, and125software systems are measured and analyzed, and the measuring results are accurate.With the further improvement of the analytical tools, this thesis realized weighted software network’s analysis. Based on this, weighted software network can describe the inner software structure more really. After the research (this is according to the method of unweighted software network), find that there also are three kinds of node in weighted software network. Then put forward three measurement formulas. These three formulas are used in the following part, and125software systems are measured and analyzed, and the measuring results are accurate. After contrast, find that weighted ripple degree is more accurate than unweighted ripple degree on the measure of software structure’s ripple effect. While, for procedure-oriented software, we still have to use unweighted ripple degree to measure.