A Method For Predicting Performance Ranking Of Features To Be Developed Based On User Feedback

In recent years,the mobile internet has developed rapidly.With the widespread application of 4G networks and the deployment of 5G networks,mobile applications(apps for short)have achieved rapid development and will be in the growth stage for a long time.The mobile application market is flooded with developers.Compared with traditional software,mobile applications have the characteristics of fast iteration and shorter update cycles,especially the more popular mobile applications.This is a new challenge for developers.The Android system has the largest number of mobile users in the world,occupying the absolute first place in mobile system share.Unlike the closed source of the Apple mobile application market,the Android market has a highly open and highly free environment,attracting a large number of users and developers,making the Android application market highly prosperous.Richer and more numerous.While developers have flooded into the Android mobile application market,the development of apps has attracted widespread attention from software engineering researchers.At present,the research work mainly focuses on tool development,application maintenance,mobile security and user comment feedback in the mobile application market.In the process of software rapid iterative update and software development,in addition to the features that users urgently require and the necessary features of products,developers will face many features to be developed,including the features proposed by developers and new features proposed by users.Faced with such diverse user requirements,limited to time and resource costs,it is difficult to implement them all in one iteration.Faced with such a variety of user needs,limited to time and resource costs,it is difficult to achieve all of them in one iteration.In the face of the above problems,this paper proposes a method to predict the performance ranking of to-be-developed features based on user feedback,which provides guidance and Suggestions for developers by analyzing the performance of to-be-developed features in existing products in the market.To be specific: first,select App products in the same field as the products to be developed,extract features and user feedback from their introduction by natural language processing technology,and define a feature set to value(FSV)information model to formalize features and user feedback information.Secondly,the Word vector space can associate feature sets with the models.Finally,Utilize the predetermined scoring rules,and assign the corresponding value according to the result of the App associated with the feature to be developed.In order to verify the effectiveness of the method,we collected a large number of app-related data from the Google Play platform,verified the reliability of feature extraction by calculating the recall rate and accuracy,and verified the effectiveness of the method with a reasonable and scientific questionnaire.Experimental results show that the methods proposed in this paper can effectively provide reference Suggestions for developers to choose which features to develop first.