Empirical Study On Underlying Relationship Between User Requirements And Edition Transition Of Mobile Applications

With the flourish of smart phones and Internet,millions of mobile applications,which are released on the Application Store,could be installed on users' phones and provide service for the users.Users use the Apps by downloading them from Application Store,at the same time they give feedback about the quality of Apps to the developers by leaving reviews which are very important for the developers on Application Store.When developer update their Apps,users' needs and feels are as important as the development of the Apps.Incorporating user reviews into iterative delivery of new App versions would improve quality and ratings of Apps.Until now there is no explicit answer on whether and to what degree App developers make use of user reviews sufficiently and timely.To deal with the problem,We extract requested features in user r eviews and updated features in new versions,identify the latent relation between them,and find 7 types of Update Patterns(UPs)by grouping similar Atomic Update Units(AUs).UPs delineate common behavioral characteristics of acting on user reviews from perspectives of feature intensity trend,sufficiency and responsiveness.Statistics are conducted to explore the similarity/difference between exhibited update patterns w.r.t.Apps,features,and time.Results would help developers get a clear understanding on their own habits of how to act on user reviews,and thus offers suggestions on utilizing user reviews more efficiently in App development.The content of this paper includes four parts primarily(1)An automated data collected tool which is designed for the update logs and reviews on Google Play is developed.The disposition and maintenance of the tool is introduced,too.We define the target data model and analyze the enormous simulation data that we collected.Then,we get some conclusions in the sta tistical sense.(2)We define the Atomic Update Unit(AUs)and introduces the generating method for it.Focusing on AUs,A method is given for calculating timeliness,sufficiency,intensity of feature update and intensity of feature request.A segment fit ting algorithm to normalize the Intensity Trend Chart for Feature Request and Update(TC)is created.(3)We define the identify the Update Patterns(UPs).We observes the latent relation between user reviews and App update logs and find 7 types of Update Patterns by grouping similar Atomic Update Units(AUs).(4)A series empirical analysis are carried out and we get some conclusions.Which type of the UPs will the developers choose when they update their Apps depends largely on individual habits rather than the nature of the features.There exists polarity between the steadily UPs choosing habit and the unsteadily one.65 percent of the App choose an unsteadily way when using the UPs.By contrast,12 percent of the App,which always shows the same type of UPs when updating,choose a very steadily way.There is a positive correlation between two types of UPs and the number of reviews that the App receives.There is a negative correlation between three types of UPs and the rakings of the App.There is a positive correlation between one type of UPs and the scores of the App.