A game recommender system using collaborative filtering (GAMBIT)

We present a recommender system anticipated to be used by community of gamers. This system filters out or evaluate games through the opinions of other similar gamers using collaborative filtering technique and suggest those to the intended user. The system uses individual ratings given by the members of community, along-side rating of the games that a particular gamer likes, in order to predict and recommend new games to that gamer. The aim is to recommend games that match the user preferences and then user-based collaborative filtering is applied on the individual ratings of the games for a particular gamer to find similarity between those gamers. Genre-based filter is smeared on the shared rated games. The prediction algorithm is also tested on a larger standard dataset, Movie Lens, in order to verify the prediction accuracy. The results of our collaborative filtering approach is also compared using Mean Absolute Error. A working web based system is presented that found high user satisfaction in terms of usability and recommendation quality.