The Research And Implementation Of Automated Evaluation Model For Performance Test Scripts

With the rapid development of software technology,the competition between software products is increasing,and the requirements on software quality are getting higher and higher.Therefore,in the software life cycle,software testing has become increasingly important.In order to ensure the stability and efficiency of software,performance testing is an important part of software testing and is a skill that every developer needs to master.In recent years,online education has grown rapidly,and many students learn knowledge about performance test through online courses.In the performance test online courses,evaluating students' performance test scripts requires much time for teachers and teaching assistants.Therefore,in order to reduce the burden,it is necessary to study the automated evaluation of performance test scripts.This thesis evaluates performance test scripts by measuring the test adequacy.At present,there are many methods to measure test adequacy,such as the control-flow adequacy criteria,mutation testing,etc.However,no one has proposed a method to measure the test adequacy of performance test scripts.Therefore,by studying other measurement methods,this thesis proposes an automated evaluation models for performance test scripts.First,starting from the script structure,this thesis extracts basic test components and test methods from performance test scripts,and summarizes a set of key evaluation items.These evaluation items cover the basic test components and test methods of performance test scripts,such as transactions,parameterization,etc.According to the importance of each evaluation item in teaching,this thesis assigns a weight to each evaluation item.This set of evaluation items and weights serves as a test adequacy criteria based on the script structure and is used to measure the test adequacy of scripts and calculate scores.In addition to the script-based structural testing,this thesis studys the existing fault-based testing,and uses mutation testing and the mutation-kill rate as the other method to evaluate scripts.Based on performance defects,this thesis proposes a set of mutation operators to generate mutants that contain performance defects.This thesis applies these two test adequacy measurements to the automated evaluation model.The model uses two evaluation methods:the static evaluation method and the dynamic evaluation method.The static evaluation method uses static scanning techniques and does not care about the dynamic execution results.By statically scanning performance test scripts,the model extracts test components and their configurations,and compares configurations with the standard rules.The dynamic evaluation method uses execution results of the script for evaluation.According to the system interacting with the test script,there are two kinds of dynamic evaluation methods:one is to interact with the system under test,and the script is evaluated by the dynamic execution results;the other is to interact with multiple mutants,and the script is evaluated by the mutation-kill rate.Based on the results of the two evaluation methods,the model calculates the weighted mean as the final score of the script.Based on the automated evaluation model,this thesis implements an automated evaluation tool on the JMeter.In order to evaluate the accuracy of automatic-evaluated scores,this thesis organizes a performance test for over 200 students and used 2 methods to score student scripts:automated scoring and peer review.By comparing and analyzing the differences between the two scores,the experiment shows that the automated evaluation score is more accurate than peer review's,which indicates that this model is reliable.Besides,this thesis also analyzes the errors in these scripts,which shows that the weights of evaluation items in the model are of educational significance.Nowaday,no automated evaluation method for performance test scripts has been proposed.Therefore,the evaluation model proposed in this thesis is highly innovative.The experiment shows that the model can effectively evaluate the correctness and design of performance test scripts,improve the efficiency and accuracy of evaluation and greatly reduce the time.The evaluation model serves as a tool for students' practices and exams and can be used in online and offline courses.