MMORPG,because of its particular characteristics,has presented interesting sociality and community.Studying game log data can not only analyze and optimize the game settings and design,but also apply the behavior comprehension from the virtual game player to real-life human behavior.Existing methods typically are based on individual players or overall data.There is no desirable approach to study the evolution of the population and the characteristics in the game.We firstly collect the game data and analyze the statistical attributes and some interesting features.To take a step further,this paper presents a novel visual representation,called category-evolutional river map,which illustrates the virtual attributes of the game players' daily changes.High-dimensional attributes are abstracted and clustered to provide the genres of the player groups,and we employ a high-dimensional data visualization method to show it.We design and implement a suite of interactive and auxiliary views,easing the exploration and analysis of the gameplay.This dataset set comes mainly from a popular online game,which is collected in 18 months with nearly 1.5 million players.In order to verify the method and system availability,we also care-fully design a number of case studies,whose results show that our system is of great significance in studying the player groups' behavior. |