Research And Realization Of Improving The Accuracy Of Online Game Player Classsification By Process Mining

With the the demand of people for the game growing,Some games have already begun to affect people’s living habits and consumption habits.With the increasing number of game players,traditional games player classification system can not meet the increasing needs of game operators.Therefore,how to distinguish different kinds of game players and guide their behavior better to bring the benefits of becoming a serious problem for game operators.Since most game has complex log structure,the traditional online game players classification system directly use clustering algorithm directly,which unable to meet the growing needs of game operators.Most game players have their own gaming habits,the same kind of players’ gaming habits is quite similar.The current number of scholars began to construct the game player behavior to behavioral model for player classification.The paper introduces the process mining techniques to build player process behavioral models by the player behavior and used them in network game player classification in order to improve the accuracy of the game player classification.However,the traditional process mining technique for unstructured log data of the game is not very good.Aiming at solve the above problems,in the original α algorithm proposed F-α algorithm,to construct game player behavior as process model with unstructured log.Then apply cluster algorithm to player classification system with the models.the specific contents are as follows:1.Describe the research status of process mining technology and applications,which focuses on the unstructure process.2.Proposing F-α algorithm on the basis of α algorithm,through the sub-division process,the polymerization of activity and abstract it to clear the process model,and analyzed the experiment results.3.Conbining process mining with clustering algorithm improve the accuracy of game player classification,and analyzed the experiment results.