Research On The Practical Application Of Code Test System Supporting Test Oracle

Unit testing is the basis of software testing,a lot of problems can be early found,and the cost of modifying the fault is lower.Automation of unit tests enables a dramatic increase in software quality at a lower cost.Test oracle is an integral part of unit testing,but also the weak link in software testing research.Test oracle is a mechanism to determine whether a program under test is performing as expected at a given test input.Automated test oracle not only effectively reduce the burden on testers,but also provide uninterrupted continuous testing to provide strong support.In the practical application of code test system(CTS),the test result evaluation module uses manual test oracle.It relies on the testers to manually calculate the expected value,manually compare the expected value with the actual value,and manually judge whether the test case passes or not.This method has obvious shortcomings in terms of efficiency and accuracy.The code test system used small programs to validate each module before,but did not use actual large projects for practical experiments.The work of this paper is mainly divided into two aspects,on the one hand,the code test system for practical research;on the other hand,the test oracle part of the automated generation technology research.Around the above two aspects of the following work:1.Propose a test oracle generation method based on neural network.In this method,the neural network is trained by selecting the appropriate neural network type and using the correct historical data to obtain the oracle information generator.With the oracle message generator,the expected value of the new input data can be predicted.2.Through the modeling and simulation of Back Propagation(BP)neural network,probabilistic neural network and generalized regression neural network,the advantages and disadvantages of these three commonly used neural network models are analyzed.Based on this,the relationship between probabilistic neural network and generalized regression neural network Combination of prophecy generating strategy.Based on this strategy,the test oracle module is programmed and validated.3.Practical research on CTS is carried out through practical engineering.CTS practical experimental results show that,CTS has good practicality.The effectiveness of the oracle module experimental results show that the proposed method of partially automated oracle generation has a good performance in the classification problem,the error between the predicted value and the actual value is only 0.1.