GameNet is a search front-end to an NLP model of game descriptions from popular web sites. It takes game titles as input and then ranks results based on relatability and difference from that input. Each search result page lists the 50 most related and least related titles, and provides links to each, allowing users to click around the model. Each seach page also provides default searches for game videos in YouTube, images in Google Image Search, and, if available, the source document used to derive each game in the model. The “ontology” model is based on Wikipedia, while the “gameplay” model is based on second LSA derived corpus of GameFAQs walkthroughs. This is a collaboration with James Ryan at the Expressive Intelligence Studio.

Gamenet is live at this link. For more information on the model and its evaluation, please check out my publications. Below are some pictures of the interface.