Applied Research of Unsupervised Machine Learning in Game Anti-fraud

doi:10.3969/j.issn.1671-1122.2020.09.007

Abstract

Abstract:

As the online game market continues to grow, there are more and more events of "get a deal" happen in the online game, which has had a serious impact on the balance of game assets, especially the interests of game publishers. This paper proposed a game bot detection method based on unsupervised machine learning, this method focused on discovering the differences in behavior between game bots and human players, introduced the word2vec idea to process the event type vector, discovered game bots and new fraud patterns through cluster analysis. After applied unsupervised machine learning to the online game anti-fraud engine, the accuracy of online game bot detection increased by about 8%, greatly improve the detection accuracy rate.

Key words: unsupervised machine learning, time series, game bot, game anti-fraud

CLC Number:

TP309

XU Yu, ZHOU You, LIN Lu, ZHANG Cong. Applied Research of Unsupervised Machine Learning in Game Anti-fraud[J]. Netinfo Security, 2020, 20(9): 32-36.

Figures/Tables 9

References 11

[1]	KANG A R, WOO J, PARK J, et al. Online Game Bot Detection Based on Party-play Log Analysis[J]. Computers & Mathematics With Applications, 2013,65(9):1384-1395.
[2]	XUE Hongyan, CHEN Ruoping. Strengthen Marketing Offenses and Defenses to Completely Contain the "Wool Party"[J]. Internet Economy, 2017,17(12):26-31.
	薛洪言, 陈若平. 加强营销攻防全面围堵“羊毛党”[J]. 互联网经济, 2017,17(12):26-31.
[3]	HAO Guanghao. Application of Digital Fraud and Fintech Anti-fraud[J]. Taxation and Economy, 2019,19(6):40-47.
	郝光昊. 数字化欺诈与金融科技反欺诈的应用[J]. 税务与经济, 2019,19(6):40-47.
[4]	CHEN Sheng, ZHU Guosheng, QI Xiaoyun, et al. Custom User Abnormal Behavior Detection Based on Depth Neural Network[J]. Computer Science, 2019,19(S2):442-445.
	陈胜, 朱国胜, 祁小云, 等. 基于深度神经网络的自定义用户异常行为检测[J]. 计算机科学, 2019,19(S2):442-445.
[5]	PING Ping, LI Jianhua, MAO Yingchi, et al. Image Encryption Based on Chaotic Mapping and Bit Reorganization[J]. Journal of Image and Graphics, 2017,22(10):1348-1355.
	平萍, 李健华, 毛莺池, 等. 混沌映射与比特重组的图像加密[J]. 中国图象图形学报, 2017,22(10):1348-1355.
[6]	WU Baogui. Exploration of the Ways to Apply Artificial Intelligence to the Field of Financial Supervision[J]. Modern Economic Information, 2019,19(18):273-274.
	武宝贵. 人工智能应用于金融监管领域的途径探索[J]. 现代经济信息, 2019,19(18):273-274.
[7]	DataVisor. Using Automated AI Technology to Combat Intelligent Network Fraud[J]. Security & Informatization, 2019,19(10):42.
	维择科技. 运用自动化AI技术打击智能化网络欺诈[J]. 网络安全和信息化, 2019,19(10):42.
[8]	TAO J, XU J, GONG L, et al. NGUARD: A Game Bot Detection Framework for NetEase MMORPGs[C]// ACM. The 24th ACM SIGKDD International Conference, August 19-23, 2018, London, United Kingdom. New York: ACM, 2018: 811-820.
[9]	KEOGH E, RATANAMAHATANA C A. Exact Indexing of Dynamic Time Warping[J]. Knowledge and Information Systems, 2005,7(3):358-386.
[10]	MIKOLOV T, CHEN K, CORRADO G, et al. Efficient Estimation of Word Representations in Vector Space[EB/OL]. https://arxiv.org/abs/1301.3781, 2013-6-15.
[11]	LE Q V, MIKOLOV T. Distributed Representations of Sentences and Documents[C]// IMLS. 31st International Conference on Machine Learning, June 21-26, 2014, Beijing,China. MA: MIT Press, 2014: 1188-1196.

	非向量化			向量化
特征	准确率	召回率	误封率	准确率	召回率	误封率
游戏切换事件时间序列	60.70%	39.00%	1.01%	62.70%	44.75%	1.07%
金币变化事件时间序列	59.49%	37.63%	1.03%	62.50%	41.88%	1.01%
点击事件变化时间序列	77.78%	66.50%	0.76%	80.99%	69.75%	0.66%
金币值变化时间序列	59.39%	31.63%	0.87%
游戏时长变化时间序列	85.73%	75.88%	0.51%

特征	准确率	召回率	误封率
统计特征+短文本特征	84.93%	88.75%	0.63%
向量化点击事件变化时间序列特征	80.99%	69.75%	0.66%
游戏时长变化时间序列特征	85.73%	75.88%	0.51%
统计特征+短文本特征+向量化点击事件变化时间序列特征	81.48%	95.13%	0.87%
统计特征+短文本特征+游戏时长变化时间序列特征	85.16%	90.38%	0.63%