Abstract: In order to overcome the problems of low data reliability and long generation time of traditional automatic generation methods of software test data, an automatic generation method of software test data based on improved Markov model is designed. Firstly, collect software test data in different stages; Then, by calculating the similarity of the collected software test data, remove the test data with high similarity, calculate the importance of the software test data with the help of entropy weight method, and complete the data preprocessing; Finally, the Markov model is improved with the help of genetic algorithm, generation path and variation factor of software test data are set, and the improved Markov model is used to automatically generate high quality software test data. Experimental results show that when the number of experiments is 50, the generation time of this method is about 2.8 s, the reliability coefficient is always higher than 0.8.
