Earticle

최근 온라인 게임 시장 규모가 급격히 성장하면서, 게임 내 재화를 획득하기 위한 부정행위가 빈번하게 발생하고 있 다. 대표적인 부정행위 중 하나인 게임 봇(game bot)은 게임 내 재화를 부정하게 수집하여, 게임 내 균형을 파괴하 고 콘텐츠를 빠르게 고갈시켜 게임 수명을 단축시키는 문제를 야기한다. 본 논문에서는 사용자 행위 로그를 입력으 로 하는 다층 퍼셉트론(Multi Layer Perceptron)을 적용하여 정상 사용자와 게임 봇을 분류하고, 각 행위가 분류 에 영향을 끼친 정도를 수치화하여 판단 근거를 추론하는 모델을 제안한다. 제안한 모델을 ‘AION’ 게임의 실제 로 그 데이터에 적용하여 10겹 교차 검증으로 테스트한 결과 약 98.4%의 정확도와 99.6%의 재현율을 보였다.

As the online game market has grown rapidly in recent years, cheating has frequently occurred to get items in the game. Game bots, which are one of the most representative cheating behaviors, collect items in the game unfairly, causing problems in the game by destroying the balance in the game and rapidly depleting the content to shorten the game life. In this paper, we propose a model that classifies normal users and game bots by applying a Multi-Layer Perceptron(MLP) with user action log as input, and infers the basis of judgment by quantifying the degree to which each action affects the classification. The proposed model was applied to the actual log data of the ‘AION’ game and tested with 10-fold cross-validation, showing an accuracy of about 98.4% and a recall of about 99.6%.

In online game, Game-bot is a kind of AI based software which performs laborious tasks for a game-player to accumulate resources such as experience, in-game items. The game-bot causes serious problems of balancing and other game-players’ play in the game. So, detecting and avoiding game-bot’s behaviors in online game is one of the important issues in live game management. This paper proposes a study on a player reaction in MMORPG environment focused on two games of 「TERA」 and 「Knight Online」. We first characterize core game-elements to affect a player’s reaction in game environment change and finally propose a template of a player’s reaction in the game environment change. The paper will contribute to detect a behavior of a game-bot in online-game, effectively.

온라인 게임 시장이 성장하면서 게임 봇의 사용은 게임 서비스에 가장 심각한 문제를 야기하고 있다. 본 논문에 서는 MMORPG 장르의 게임 봇 중 채집을 진행하는 봇을 탐지하기 위한 채집 좌표 분석 모델을 제안한다. 제안한 모델은 좌표 데이터를 기반으로 플레이어의 채집 행위를 분석 한다. 정상적인 플레이어보다 손쉽게 게임 내 재화와 아이템을 수급할 수 있는 게임 봇은 수면 시간, 캐릭터 조작 피로도와 같은 현실적인 제약의 영향을 받지 않기 때문에 채집 행위를 시도하는 좌표 구역에 차이가 발생한다. 좌표 구역을 나누고 각 플레이어의 좌표 구역 차이를 이용하여 게임 봇 플레이어와 정상적인 플레이어를 구분해 낼 수 있도록 했다. NCSoft 사의 AION 로그로 데이터셋을 만들고 random forest 모델에 적용하여 게임 봇을 탐지한 결과 재현율 72%, 정밀도 92%의 성능을 보였다.

As the online game market grows, the use of game bots is causing the most serious problem for game services. We propose a harvest coordinate analysis model to detect harvesting bots among game bots of the Massively Multiplayer Online Role-Playing Games(MMORPGs) genre. The proposed model analyzes the player's harvesting behavior using the coordinate data. Game bots can obtain in-game goods and items more easily than normal players and are not affected by realistic restrictions such as sleep time and character manipulation fatigue. As a result, there is a difference in harvesting coordinates between normal players and game bots. We divided the coordinate zones and used these coordinate zone differences to distinguish between game bot players and normal players. We created a dataset with NCSoft's AION log and applied it to a random forest model to detect game bots, and as a result, we derived performance with a recall of 0.72 and a precision of 0.92.

An approach for game bot detection in massively multiplayer online role-playing games (MMORPGs) based on the analysis of game playing behavior is proposed. Since MMORPGs are large-scale games, users can play in various ways. This variety in playing behavior makes it hard to detect game bots based on play behaviors. To cope with this problem, the proposed approach observes game playing behaviors of users and groups them by their behavioral similarities. Then, it develops a local bot detection model for each player group. Since the locally optimized models can more accurately detect game bots within each player group, the combination of those models brings about overall improvement. Behavioral features are selected and developed to accurately detect game bots with the low resolution data, considering common aspects of MMORPG playing. Through the experiment with the real data from a game currently in service, it is shown that the proposed local model approach yields more accurate results.

Game bots are illegal programs that facilitate account growth and goods acquisition through continuous and automatic play. Early detection is required to minimize the damage caused by evolving game bots. In this study, we propose a game bot detection method based on action time intervals (ATIs). We observe the actions of the bots in a game and identify the most frequently occurring actions. We extract the frequency, ATI average, and ATI standard deviation for each identified action, which is to used as machine learning features. Furthermore, we measure the performance using actual logs of the Aion game to verify the validity of the proposed method. The accuracy and precision of the proposed method are 97% and 100%, respectively. Results show that the game bots can be detected early because the proposed method performs well using only data from a single day, which shows similar performance with those proposed in a previous study using the same dataset. The detection performance of the model is maintained even after 2 months of training without any revision process.