A Research On Trust-Based Crowdsourced Collaborative Testing And Algorithm Implementation

The conception of crowdsourcing means an idea of outsourcing a task that is traditionally performed by an employee to a large group of people on the Internet in the form of free and voluntary. With the popularity of the Internet and Web 2.0 technologies, crowdsourcing has grabbed people’s attention as a new model of distributed problem solving and commercial production. Web based applications are increasingly becoming more feature rich, important and also the most popular means for developing commercial systems in catering to large number of end users spread across the globe. Testing Web applications in a crowdsourced environment is becoming one appealing solution because of its possible economical cost and exposure to a wide audience.A variety of approaches have been proposed to address collaborative open-source testing and GUI testing in the software engineering domain. However, less is known about the quality control of crowdsource testing in an open environment like Internet, especially the appropriate assignment of test cases to appropriate testers. To address the issue of testers’trustiness, this thesis proposes a trust-based collaborative crowdsourced Collaborative Testing. In doing so, this thesis summarizes and classifies the state-of-art techniques of crowdsourcing and the mainstream crowdsourcing platforms. Also we illustrate the available crowdsourced collaborative testing approach and its algorithm implementation. Finally we introduce dynamic trust management into crowdsourcing by using a computational model of trust based on probability and statistics of testers’ historical performance. The model first figures out the objective trust by counting the number of and level of bugs, found so far in the testing procedure. Also the subjective trust is obtained by evaluating the truthfulness and description of the bugs. Then the overall trust is obtained by integrating the objective and subjective trust. Experiments have demonstrated that the present model can accurately reflect the trust of crowdsourcing testers and improve the efficiency of testing to some extent.