Analyzing Comments Made to The Duolingo Mobile Application with Topic Modeling

Abstract

There are approximately 6.4 billion mobile users in today's world. With this widespread use, people now prefer devices that are easier to carry and use, such as mobile devices. At this point, mobile applications play a very important role. Mobile applications are preferred for many purposes, especially for entertainment, education and social activities. In foreign language learning, there are hundreds of mobile applications that contain different activities for many foreign languages. Duolingo is an application that stands out in terms of providing many foreign language support and learning by gamification. Users of this application, which has such a widespread use, give feedback with their ratings and comments. More than one million comments have been left on the Duolingo app so far. This study was carried out in order to analyze these comments and to determine which points the users bring to the fore, and to provide preliminary information for both software developers and those who want to look for similar applications based on these points. The comments were analyzed by the Latent Dirichlet Allocation (LDA) method using the Python programming language and NLTK, Gensim libraries. The topics detected by both the word bag term frequency and the term frequency-inverse document frequency (TF-IDF) method are shown.