Abstract

This paper proposes a new framework for automated
analysis of game-play metrics for aiding game designers
in finding out the critical aspects of the game caused
by factors like design modications, change in playing
style, etc. The core of the algorithm measures similarity
between spatial distribution of user generated in-game
events and automatically ranks them in order of importance. The feasibility of the method is demonstrated on
a data set collected from a modern, multiplayer First
Person Shooter, together with application examples of
its use. The proposed framework can be used to accompany traditional testing tools and make the game design
process more efficient.