Research On Modeling And Assessment Methods For Benignization Of Software Dynamic Evolution

Software quality has great influence on people’s work efficiency and the quality of life.The study of the benign degree of software are very important.The modeling and assessment method for benignization of software dynamic evolution is studied.The problems addressed in the paper are as follows:1. Benign degree model of software dymamic evolution is constructed.Continuous operation of software under open enviroment suffers from ever-changing user requirements and operating enviroment,combined with the deficinecy of predefine evolution rules under enclosed enviroment leads us to present a run-time model based benign degree evolution framework for software dynamic evolution to attack unpredictable context events under open enviroment.The run-time model inclubs context model,demand model,strategy model and architecture model etc.If the predifined evolution operation can’t solve the unpredictable,the model’s benign is guided by model based the planning of evolution to meet user’s high-level requirement.2. Software dynamic evolution benign assessment. In this paper, the number of gray system theory in the number of gray to represent the value of software metrics benign dynamic evolution model. Then according to the principle of unity of the benign effect on the original metric pretreatment to collect qualitative and quantitative metrics value of the data into the utility value. Then get the ideal benign assessment of the positive and negative values by seeking the ideal point method, using the Minkowski distance to assess the degree of software virtuous cycle.Finally, K-Means clustering algorithm to assess the results of the dynamic evolution of integrated software benign degrees.3. Experimental verification software benign modeling and assessment methods. In this paper, the degree of model-based software benign, indoor environmental comfort temperature control software for acquisition and processing benign index, and then through Matlab simulation data. Finally, the results were compared with other algorithmsby evaluating the proposed method of simulation results to verify the validity of the proposed method.