| Mobile app bug report describes problems with app,such as a crash,an unexpected result or something not working as designed.With the rapid development of mobile app market,those bugs of app may damage the benifit of a lot of users.In order to improve the quality of mobile app,this paper proposed a system to analysis app bug reports.Firstly,we give a summary of text classifying and text clustering technologies and compare the algorithms of these two methods.Text clustering is about partitioning data set into groups so that the data objects in the same group are more similar to another.Clustering can make developers acknowledge the information of bug reports more quickly and convenient.Besides,this paper introduces some methods and metrics of text preprocessing.Secondly,we propose a clustering system which implements a semi-supervised clustering algorithm "Cop-KMeans" to analysis bug reports.The system not only clusters bug reports,but also collects labels of the result of clustering.With those labeled data,the system could achieve a higher accuracy of clustering.Finally,we use the bug reports from KiBug which is a mobile app testing platform to evaluate the performance of "Cop-KMeans".The experiment result shows the algorithm can correctly cluster the bug reports.This paper also makes the detail experimental evaluation with "Cop-KMeans" in the aspect of performance and accuracy. |
PDF Full Text Request