Text Analysis of Video Game Reviews
INFX 575 Spring 2015 Final Project Group
Abstract
Video game have reached a saturation point where there are devoted media sites to reviewing video games as well as news around the gaming industry. However, there has not been research done on game reviews the same way that other entertainment industries like movies and music have. We seek to find textual patterns and characteristics in video game reviews by scraping established video game sites and conducting latent dirichlet allocation, jargon distance, and sentiment analysis to identify common writer characteristics and habits. We found that most writers typically score games between 60 to 70 out of 100 and use similar phrases to describe those games in their reviews. Further, we found through sentiment analysis that most writers use positively associated words, even for games that score poorly. We suggest more detailed analysis for games in certain genres to see if these results are consistent within a subpopulation of the reviews and propose suggestions for future work which include a write recommendation system.
Authors and Contributors
Fei Guo Victor Li Jonathan Lin Xinyu Zheng