A Recommender System for an Online Game Platform in the Metaverse

Recommending new content based on a users’ history has become a staple in many areas of the internet—from e-commerce to news websites. These recommendation strategies use various techniques that involve data mining what content a user has viewed in the past to suggest new content. In this paper, we present approaches to a recommender system for the digital video game platform called Steam. In particular, we design three approaches to recommending games in the metaverse. One approach makes predictions using information about what games a user has played in the past. Two other approaches utilize user-generated tags to get targeted results that are more suited to the users’ preference. Evaluation results demonstrate the practicality of our approaches in utilizing user-generated tags weighted by playtime to generate recommendations, allows for more niche games that may have been initially overlooked.