New Software Testing Technology Research And Implementation

Software test is the main measure to guarantee the software quality. The design of test data is very important, key is the automatically generation of test data. As the software technology develops, application becomes bigger and more complicated. It is very low efficient to test and analyze by man, which could not guarantee the quality. To generate a large scale involved test data is the main problem, besides, those data should belong to different classes. Stability of software could be greatly improved and many human resources could be saved if there is a tool which can analyze application and generate test data automatically. There are many automatically test tools at present, but most of them have faults, cannot generate test data without human, and they can deal with limited types of data. Often, they're used for unit test.First, this paper described the development of automatically testing tools and the basic theory of software automation. Then dived into Genetic Algorithm and Markov, and put forward some improvement.This paper analyzed the self-fit algorithms based on Genetic Algorithm, and put forward improved self-fit GA.This paper introduced the combination of Genetic Algorithm and simulation Genetic Simulated Annealing Algorithm verified its advantage on best search.Also, this paper put forward a new test method which was path-oriented. Markov used random procedure model to describe how to use software. This paper generated test cases based on randomized procedure. Markov model also can use Genetic Simulated Annealing. Finally, this paper combined Markov and Genetic Algorithm, created the prototype of automatically test tools.