The Design And Implementation Of Crowdsourced Feedback Review And Delivery System Based On Test Report Summary

Crowdsourced testing involves hiring a large number of online crowd workers to test application and give feedback in a short period of time.Over the past few years,crowdsourced testing has gradually become a popular method for software testing as it brings abundant test conditions and real user scenarios,which also has a relatively low labor cost.However,unlike the traditional testing,crowd workers are consisted of a large amount of people,mostly non-professional testers,which leads to large quantities,high repetition rates and low quality of the test reports.It is difficult for the reviewers to inspect numberous and repetitive test reports in a short period of time and deliver feedback in a timely manner.In order to improve business processes and reduce the difficulty of the re-view,the crowdsourced test platform requires a report review and delivery system,which is specified for crowdsourced testing features to assist the work.By utilizing the crowdsourced test platform,this thesis has designed and implemented a crowdsourced review and delivery system based on the test report summary.The system has introduced a novel mechanism,which is based on reporting summarization to solve the problems of the crowdsourced feedback review.First,the system clustered reports based on the features of text and screenshots,thereby aggregating similar reports into the same cluster.Second,based on the PageRank algorithm,the system weighted reports from each cluster,and then the one which had the highest weighted value from each cluster was chosed to be main report of the cluster.Third,the system splited reports into sentences and screenshots from each cluster,which were then compared with the main report to find the different points.Last,the system clustered the different points.Each cluster served as a supplementary point to the main report.Through the identification and organization of the main report and supplementary points,the report information is enriched and more hierarchical,so that the quality of reports is improved.The system was divided into review and delivery module,text process module,image process module,aggregation module and summary module.The system has made full use of load balancing,out-of-process cache Redis.etc.,to achieve server statelessness,which has allowed it to expand horizontally.Moreover,making full use of asynchronous and event-based design has improved system loose coupling.Master-slave synchronization has been used to ensure data reliability.Currently,this system has been launched on the testing platform to support the crowdsourcing business of the company.The system has effectively aggregated similar reports and reasonably organized redundant information,which enables the reviewers to identify the helpful information more quickly and accurately.Finally,the system has im-proved the efficiency of the report review,guaranteed the quality of the report delivery and provided a better delivery of the test product to the task issuer.