Length and sentiment analysis of reviews about top-level video game genres on the steam platform

Abstract

During game development, a common uncertainty revolves around the emotional reactions players might exhibit when experiencing newly released games. Understanding emotions and experiences of video game players is complex. As a first step, players can share their experiences of video games by writing reviews. By investigating these reviews, the emotions, experiences, concerns, and opinions of players can be understood. The goal of this study is to provide a deeper insight for video game developers to understand the players' emotional reactions regarding the top-level genres. Besides, as subgoals, it can be observed whether a connection exists between word numbers and review type (positive/negative); whether there is a difference between the type of emotions in the case of different genres; or whether the emotional valence changes during writing a review. In this paper, overall 35, 983, 481 reviews of 11 top-level video game genres are studied using natural language processing methods with the NRC Emotion Lexicon in the statistical program package R. The results show that people write negative reviews earlier than positive ones and no correlation exists between the time at review and their word number. A connection can also be observed between word numbers and whether a review is positive or negative: the median review length is 40 words in case of negative reviews, while it is 19 words in case of positive ones. Regarding emotions, the emotional valence as well as sentiments’ number and percentage differ between the top-level genres. Each has its unique emotional valence and evokes various emotions.