Web Collaborative Testing Based On Crowdsourcing

The importance of Website testing is gradually recognized by everyone,people pay attention to development at the same time also began to focus on testing.However,the skills and working attitude of the testers are uncertain.The choice of whom will participate in the test will directly affect the quality of the test results.Therefore,selecting the staff with relatively high level of skills and good working attitude will be our issues that need resolving.The crowdsourcing market has become very popular among researchers as a way to use cheap workers to collect data on the Internet.This study explores how to effectively conduct Web collaboration testing in crowdsourced environments.We use the improved Dual Task Assigner(DTA)algorithm to arrive at the best worker to complete the Web collaborative testing task.In this paper Django framework is used to design a WebTesters based on the xTurk crowdsourcing platform.This system can be used as a platform for testing many Websites.Dalian Maritime University Graduate School Website is took as a case to study our test.This system completed the construction of the test tasks,upload,publish and obtain results and other operations.Then tasks are post to the xTurk Website through the WebTesters system and recruit workers to complete the testing tasks.Due to the different levels of each crowdsourced worker and the different testing tasks performed,crowdsourcing assignments need to be considered.DTA algorithm is used to solve the problem of crowdsourcing task allocation,and take the results collected by WebTesters system as the input data of DTA algorithm.However,there are many shortcomings in the DTA algorithm.This paper present a matrix factorization(MF)approaches based on the accuracy of crowdsourcing workers'history tasks to predict his accuracy over new tasks.With the MF method we obtain the accuracy of crowdsourcing workers for tasks that will be completed soon,based on which accuracy the DTA algorithm is used for task assignment.Finally,This paper compare the original DTA algorithm with the improved algorithm based on the experimental results.