Understanding Mobile Game Reviews Through Sentiment Analysis: A Case Study of PUBGm

Mobile games are increasingly gaining popularity within the gaming industry due to their unique features and experiences, distinct from conventional platforms like desktops and consoles. Understanding player feedback and reviews is crucial for enhancing game services and optimizing user experiences. This paper focuses on analyzing player reviews within the realm of mobile esports. We introduce a comprehensive framework that employs topic modeling and sentiment analysis to extract insightful keywords from a vast collection of reviews. Utilizing the Latent Dirichlet Allocation algorithm, we uncover diverse topics within the reviews. Furthermore, we exploit Bidirectional Encoder Representations from Transformers (BERT) combined with a Transformer (TFM) downstream layer for precise sentiment analysis, capturing players’ sentiments towards various topics. The experiment was conducted on a dataset containing six million English reviews collected up to March 2023 for the mobile game PUBGm from Google Play. The experimental results demonstrate the framework’s proficiency in efficiently identifying player concerns and revealing significant keywords embedded in their reviews, thereby supporting mobile esports game operators to refine services and elevate the gaming experience for all players.