Application Of Multi-agent Technology In Automated Testing Of Web Systems

At present,the main points of Web systems test are basically carried out in five parts,which are function test,performance test,user interface test/usability test,compatibility test,security test and so on.It's not hard to see from here that the work required to test a Web Systems is enormous.Some of these different parts don't even have anything in common with each other,so it takes not just a powerful test computer,but different test software,and ultimately a human to do the complicated work.In recent years,distributed artificial intelligence has been widely valued and studied by many scholars,and the main research direction in this field is multi-agent systems(MAS).MAS has great robustness and flexibility,and has the social characteristics of cooperation and coordination similar to human society.With such huge advantages,MAS can be used to solve the problems such as large computation,many test items and waste of resources encountered in the current Website systems testing.The first problem to be solved when using MAS is to solve the cooperation and coordination among agents,so the above two problems also need to be solved when using MAS in Web systems testing.In view of the current situation of Web systems testing and the problems of MAS application in testing,the following work is done in this thesis.This thesis designs use case diagram according to the classical systems,this thesis proposes to use UML use case diagram to solve the problem of agent task assignment in MAS test Web systems.In this thesis,the Agent recruitment mechanism is built to solve the problem of competition and cooperation of agents in the test environment,so as to realize the multi-agent test of different functional modules in a Web systems,so as to quickly and accurately complete the test of the Web systems.This thesis absorbs the steps and regulations of Web testing in the past,and USES them as a method to improve and supplement the Agent testing mechanism.This thesis proposes to model the graphical interface of the Web systems,which is used to check the test process and results to ensure the comprehensiveness and accuracy of the test.Then the framework of automatic test is designed,and the Website systems of scientific research project management of Hubei provincial health commission,which has participated in,is partially tested,and the model of multi-agent participation is analyzed and evaluated.The experimental results show that the proposed scheme is effective.