This is the detail page of the paper
XAI in Online Games.
As the paper is under review, we don’t put many details here for the moment.
Online gaming is a multi-billion dollar industry that entertains a large, global population. Empoweringonline games with AI has made a great success, however, ignores the black-box model explainability inorder to develop and levarage AI responsibly. In this paper, we introduced and discussed the audienceand the concept of XAI (eXplainable AI) in online games. We proposed a GXAI workflow whichcombines the strong expressiveness of multi-view data sources and the clear transparency of multi-view black-box models. We presented four specific classifiers and explainers in the character portraitview, the behavior sequence view, the client image view and the social graph view. Experimentsconducted on real-world datasets for game cheating detection and player churn prediction showed theaccuracy of classification and the rationality of explanation. We have also discovered and presentednumerous interesting and valuable findings from the individual, local and global explanations. Weimplemented and deployed three practical applications including evidence and reason generation,model debugging and testing, and model compression and comparison in NetEase Games and receivedvery positive reviews from sufficient user studies. More future work is in progress since this is thefirst work that introduces XAI in online games.
todo