How can i improve my app? Classifying user reviews for software maintenance and evolution

TLDR

App stores such as Google Play and the Apple Store provide user reviews and star ratings that contain usage scenarios, bug reports, and feature requests, but the large, unstructured volume and variable quality of feedback make identifying useful information for maintenance and evolution difficult. This study introduces a taxonomy for classifying app reviews into maintenance‑ and evolution‑relevant categories and proposes an approach that combines natural language processing, text analysis, and sentiment analysis. The proposed method merges these three techniques to automatically assign reviews to the taxonomy categories. The combined approach achieves 75 % precision and 74 % recall, outperforming each technique alone, which yielded 70 % precision and 67 % recall.

Abstract

App Stores, such as Google Play or the Apple Store, allow users to provide feedback on apps by posting review comments and giving star ratings. These platforms constitute a useful electronic mean in which application developers and users can productively exchange information about apps. Previous research showed that users feedback contains usage scenarios, bug reports and feature requests, that can help app developers to accomplish software maintenance and evolution tasks. However, in the case of the most popular apps, the large amount of received feedback, its unstructured nature and varying quality can make the identification of useful user feedback a very challenging task. In this paper we present a taxonomy to classify app reviews into categories relevant to software maintenance and evolution, as well as an approach that merges three techniques: (1) Natural Language Processing, (2) Text Analysis and (3) Sentiment Analysis to automatically classify app reviews into the proposed categories. We show that the combined use of these techniques allows to achieve better results (a precision of 75% and a recall of 74%) than results obtained using each technique individually (precision of 70% and a recall of 67%).