The Integration of NLP and Topic-Modeling-Based Machine Learning Approaches for Arabic Mobile App Review Classification

App stores serve as digital distribution platforms for mobile applications for almost every software and service. They provide users with a range of applications to browse, purchase or install at no cost. Additionally, users can post reviews in terms of textual feedback and star rating. According to recent studies, text reviews provide rich information, such as enhancement requests, bug reports, user experience, and text ratings. Such data can benefit different stakeholders, especially developers, who can examine the customer needs and react through app improvement, thus increasing the app’s popularity, quality and success in the marketplace. However, the amount of app reviews is too sheer for manual categorization. Thus, an automated approach is required to support developers in analyzing app reviews. While many studies have proposed approaches using classical machine learning algorithms for English text reviews, there is limited research on applying Arabic Text Classification for user reviews in Arabic. This paper aims to leverage Arabic app reviews, natural language processing and topic modeling-based machine learning techniques to classify and analyze user feedback.