Analysis of the Correlation Between Playtime, Design, and Game Mechanics to Positive Reviews on the Fighting Games Genre using Large Language Models

Abstrak

The video game  industry is experiencing rapid growth, with the fighting game  genre remaining a favorite due to its tactical challenges and deep mechanics. This study explored the relationship between playtime, game design, and game mechanics to positive reviews from players, using Large Language Models (LLMs) for sentiment and emotion analysis. The data was collected from more than 200,000 Steam user reviews on 12  popular fighting games. The results show that the correlation between playtime and positive reviews tends to be weak, although in some titles, longer playing duration is associated with better sentiment. In terms of game design, players prefer games with fantasy settings  (92.34%), 2D graphics (94.21%), and anime visual style  (95.12%), which significantly drives positive reviews. In the mechanical aspects of the game, features such as multiple meters (93.11%), advanced blocking (93.56%), and wall boundaries (91.72%) get higher satisfaction, suggesting that the complexity and variety of mechanics can increase player engagement. LLM-based sentiment analysis  also reveals that technical factors play an important role in player perception. The most common complaints in negative reviews relate to lag, character balancing, and additional content quality (DLC).