What users want for my app - implementing a system to analyze user reviews for mobile application maintenance activities

Abstract

User reviews of mobile applications are often provided by users about their satisfaction, complaints or suggestions on the mobile apps. They form valuable and instructive feedback for app developers in the maintenance process to fix bugs, add new features, enhance the user experience of using the app, etc. However, due to the large quantity and various qualities of the user reviews, it is not realistic for developers to manually read every review to elicit useful information for application maintenance. To address this problem, the study proposes a mobile Application Review Analysis system for Maintenance (ARAM). ARAM is able to identify the useful app review information that can contribute to mobile application maintenance, extract keyphrases from review contents, and analyze users’ sentiments for the keyphrases. A new method called DoubleRank was proposed for extracting keyphrases from app reviews. By classifying and summarizing the user reviews, ARAM produces a report of users’ opinion on the extracted keyphrases and their associated reviews, which can provide useful suggestions for developers to plan maintenance activities. The generated report is accurate when comparing with the results of manual analysis.