Issue detection and prioritization based on mobile application reviews

Opinion mining for mobile application (app) reviews aims to analyze people’s comments on app stores to support software engineering activities, particularly software maintenance and evolution. One of the main challenges for software quality maintenance is promptly identifying emerging issues, e.g., bugs. However, due to a large amount of textual data, manually analyzing these comments is challenging, and machine learning-based methods have been used to automate opinion mining. This paper introduces the automatic generation of a risk matrix from user reviews to answer the following research question: how do we prioritize and treat reviews in time so that the app is competitive and guarantees the timely maintenance and evolution of the software? We present the MApp-IDEA (Monitoring App for Issue Detection and Prioritization) method to detect issues and classify the reviews in a risk matrix with prioritization levels. We present an approach that (i) automatically collects reviews, (ii) detects issues, (iii) classifies reviews in a risk matrix, and then (iv) models the temporal dynamics of issues and risks through time series to trigger alerts. We performed an empirical evaluation with 50 mobile apps and processed approximately 5 million reviews, where we detected 230,000 issues and classified them into priority levels using a risk matrix. We found that issues detected early with our approach are associated with later fix releases by developers.