Hiroshi Itsuki, Asuka Takeuchi, Atsushi Fujita, Hitoshi Matsubara Solution
Total Page:16
File Type:pdf, Size:1020Kb
Exploiting MMORPG Log Data toward Efficient RMT Player Detection Hiroshi Itsuki, Asuka Takeuchi, Atsushi Fujita, Hitoshi Matsubara Future University Hakodate, JAPAN Summary RMT Automatic detection of RMT players using log data RMT player network Ranking all players with regard to their “suspiciousness” Currency Currency Manual verification starting from rank 1 Items Items Issues of Operating MMORPG Seller Collector Earner MMORPG: Multi-player online role-playing game Real Property Problems of RMT in MMORPGs Countermeasure against unfavarable behavior currency in Imbalances virtual economy RMT: Real-money trading virtual world Direct attacks to general players Harassment (PK, robbing) Unauthorized behavior and it’s automation ↓ Manual investigation requires human effort and time Prevents new players Effective use of huge volume of log data Discourages existing players Buyer Solution Computes “suspecthood” using statistics derived from log data Provided data Target Commercial MMORPG Log data (Action log, Chat log) “Uncharted Waters Online” List of manually identified RMT players TECMO KOEI GAMES CO., LTD. Derived four types of statistics from both types of data http://global.netmarble.com/uwo/ TCC: The total number of utterances recorded Target period TAC: The total number of action records 2009/8/30~2009/9/13 AT: The amount of minutes in which at least one action is taken (in which the operators had identified RMT players) TCH: The amount of virtual currency handled in the period Acknowledgemnt (The absolute values of currency increase and decrease) Our greatest thanks to TECMO KOEI GAMES CO., LTD. Preliminary Investigation 0. Assumption 2. RMT Player Detection RMT players handle huge amount of currency Different statistics achieved the best results 15,250 players ⇒ top 1,000 players based on TCH for each type of RMT players 1. Statistical Analysis Table 2: The Number of Extracted RMT Suspects Significant differences between RMT players and others Statistics Seller Collector Earner All Table 1: Mean Values of Statistics n 10 4 15 29 Type n TAC AT TCC TCH TCH 639 452 80 639 Seller 10 254.4 55.8 1.3 12,209.2 TAC 998 889 360 998 Collector 4 745.5 138.3 46.8 23,247.0 TCC 1,000 1,000 1,000 1,000 Earner 15 106,137.8 16,772.5 11.0 13,438.7 AT 990 891 48 990 Others 971 75,008.8 3,718.0 951.0 3,111.0 TCH/TAC 72 133 419 419 Seller TCH/TCC 226 347 349 349 Collector Most of the RMT Earner TCH/AT 81 123 643 643 Other players are silent 102 Earners (15) Earners spend long time 29 at a low frequency of action 25 101 20 Total actions counts / Activity time (log scale) 15 TCH 0 3 6 9 12 15 18 21 24 TAC 10 TCC Average activity hours per day AT Sellers and collectors take TCH/TAC TCH/TCC actions less frequently Number of identied RMT Players 5 TCH/AT no effect Seller 5 10 Collector 0 Earner 0 200 400 600 800 1000 Other Top N RMT Suspects sellers (10) RMT players deal with collectors (4) a huge volume 4 10 of virtual currency Future Work Total currency handled (log scale) Explore a more effective statistics 103 especially from short-term log data 102 103 104 105 106 Apply to other game titles Log count (log scale).