Research On Dynamic Defect Prediction Method Of Component-based Software

Component-based software development represents the further development of object-oriented technology.With the increasing complexity of large-scale software systems,component-based software development method offers more advantages than object-oriented technology in terms of design,development,flexibility and reusability.In the component-based software system,different components are called and integrated with each other.The impact of the failure of different components is different because of their different functions and positions in the software.Therefore,in order to get more accurate and practical assessment results,the importance of different components and the defect propagation trend between components should be considered in the defect prediction process of component-based software system.At the same time,the architecture of component-based software system is also becoming more and more complex.As a result,it is also an urgent problem to find a solution for dynamic defect prediction to take advantage of the existing failure information of component-based software system.In order to solve the above problems,the component-based software system is taken as the study object in this paper.According to sources of components and the use object,the dynamic defect prediction method of the component-based software system is well studied.Firstly,a software defect prediction model based on PSO improved fuzzy C-means clustering is proposed.Considering that the internal components of the system originate from the independent research and development within the company,the model uses a large amount of unlabeled data collected from the internal structure of components to predict the defect.On the basis of the fuzzy theory,the model optimizes the model parameters and the distance calculation function to realize the fuzzy clustering of the defect dataset.Then,in view of the invisible problem of the internal components of the system,a method for measuring the defect probability of component nodes based on the influence factor and defect propagation is proposed,which uses complex network theory to abstract the architecture of component-based software system,and takes the components' importance into the defect location method based on program spectrum.Considering the operation result and defect propagation tendency of the connection path between the internal components nodes in the component-based system,the defect probability of the component nodes is calculated and corrected to get more reasonable prediction results.Finally,based on the research of the defect prediction model of component-based software and the defect probability measurement method of component nodes,a dynamic defect prediction method of component-based software based on multi-variable grey model is proposed,which introduces the grey correlation analysis method and the idea of sliding window to analyze the dynamic evolution characteristics of component-based software system.As a result,the method can effectively predict the trend of component failure data and the defect discovery quantity,and provide powerful guidance for the arrangement of subsequent test work.