A Recommendation Method For APP Maintenance Based On User Reviews

The app store allows users to rate and comment on downloaded mobile applications.These reviews can directly or indirectly reflect user intention.Developers can maintain and improve their application through continuous monitoring users' reviews,to better meet customer expectations.Due to the large data,unstructured nature and different quality of user reviews,reviews analysis faces great challenges.Therefore,how to dig the useful information from reviews for developers and recommend hot information to developers become one of the most important problems in software development.In order to solve the above problems,this paper introduces an efficient solution for automatically analyzing user reviews and recommending hot reviews in user reviews to developers.First of all,according to users' reviews intention,we divide reviews into sentences granularity and classify sentences into certain classifications by using natural language processing technology,such as the problem discovery,feature request and so on.And we divide the sentences that under each type of intent into different topics.Secondly,we cluster the sentences expressing the same maintenance opinions under each topic into one group.Then,we analyze the emotional distribution of user reviews to identify the importance of user reviews.Finally,we calculate the score of different maintenance opinions by considering the multi-dimensional information such as user intention and emotional.And then the priority of user maintenance opinions is divided to provide efficient and valuable opinions for developers to carry out subsequent app maintenance,version update and evolution.In order to verify the relevant theory and solutions proposed in this paper,we designed relevant experiments.And we verify our results with the changelog of the official application.The experimental results show that the proposed scheme can effectively help developers understand user requirements and solve problems raised by users and effectively provide valuable suggestions for developers to carry out subsequent APP maintenance and evolution.