Seong-Il Yang

Crime scene re-investigation: a postmortem analysis of game account stealers' behaviors

As item trading becomes more popular, users can change their game items or money into real money more easily. At the same time, hackers turn their eyes on stealing other users game items or money because it is much easier to earn money than traditional gold-farming by running game bots. Game companies provide various security measures to block accounttheft attempts, but many security measures on the user-side are disregarded by users because of lack of usability. In this study, we propose a server-side account theft detection system base on action sequence analysis to protect game users from malicious hackers. We tested this system in the real Massively Multiplayer Online Role Playing Game (MMORPG). By analyzing users full game play log, our system can find the particular action sequences of hackers with high accuracy. Also, we can trace where the victim accounts stolen money goes.

Extracting gamers' cognitive psychological features and improving performance of churn prediction from mobile games

With the continued growth of the mobile game market, many game companies aim to make money through mobile games. In this situation, knowing the tendency of gamers and predicting the churn in advance can maximize profit through effective game services. For this reason, much study has been conducted for the purpose of gamer analysis and churn prediction. However, the study was mainly conducted using surveys, bio-signals, and PC online game logs, which are likely to make detailed information. In this study, we extracted seven cognitive psychological features from the game logs of Crazy Dragon, a commercial mobile RPG game, and used these to predict the churn. In addition, we analyzed the effect of purchasing feature by comparing the churn prediction performance according to presence or absence of purchase feature. In the conclusion, we obtained higher performance when predicting the churn using cognitive psychological features than using the basic raw logs. Also, we obtained high churn prediction performance using only cognitive psychological features without purchase feature.

Optimizing game live service for mobile free-to-play games

In this paper, we propose a method for optimizing the game live service. Especially, we focus on improving user retention. Firstly, we define player churn in the game and extract features that contain the properties of the player churn from the game logs. And then we evaluate the importance of features using random forest in classification. Finally, we build association matrix between features and suggest a reward method for in-game events according to the relative importance of features and the interrelationships between features. Results of applying to the game in service show that the proposed scheme is enough to improve user retention.

Predicting churn in mobile free-to-play games

In the mobile game industry, Free-to-Play games are dominantly released, and therefore player retention and purchases have become important issues. In this paper, we propose a game player model for predicting when players will leave a game. Firstly, we define player churn in the game and extract features that contain the properties of the player churn from the player logs. And then we tackle the problem of imbalanced datasets. Finally, we exploit classification algorithms from machine learning and evaluate the performance of the proposed prediction model using cross-validation. Experimental results show that the proposed model has high accuracy enough to predict churn for real-world application.