| With the development of Internet and web technology, new e-commerce and mobile e-commerce become more popular. Recently, the mode Online To Offline developed a lot in China. O2 O is a business model which combine offline trade and internet, online platform push lower price or some other discount massage of offline shops to users, and if one service choosen, user has to pay online and get a payment voucher, then use the voucher to the offline shop and enjoy the service. That is, O2 O platform is a assistant of offline shop reception, which can help push offline information to online. At present, anticheat work mostly based on manual work. offline department monitoring if the amount of a shop’s consumption are abnormal, etc., and then check offline. However, the current O2 O account highly relies on the phone number of the user, but the cost of replacing the phone number is very low, resulting the cheater can easily replace their account by change the phone number.Our project belongs to a takeout O2 O company’s anti-cheating department. Starting from the point anti-cheat, a method that can predict user’s risk of cheat through a combination of machine learning and data mining is proposed, to reduce the monitoring cost. At the same time, due to the further operation of high risk of users, this paper also developed a corresponding background operation module. At present, anti-cheat system has been on-line for five months, and after the two version’s upgrade, data shows that this method can effectively identify high risk users, and reduce the complexity of anti-cheat departments’ work. |
