Research On Risk Management Of Ecommerce Transactions

In recent years, with the rapid development of ecommerce, online shopping became the people's major shopping mode, Ali's sales reached 91 billion 200 million in November 11,2015, and JD also reached 100 billion. Behind the prosperity, the outbreak of large of mount of transaction risk problems and the outdated transaction risk management not only became the bottleneck that hampered the healthy development of the ecommerce, but also affected ordinary customers'legal rights. Thus, the fairness, justice and authenticity of ecommerce has been challenged by the public. Their experiences in the transaction risk management are still relatively insufficient and the development of risk management started relatively late. The transaction risk assessment is still relied on the business experience. With the growth of the volume and increase of complexity in ecommerce, especially the professionalization of the development of black industry chain, the model of traditional transaction risk management was unable to meet the need of risk management development. Therefore, to construct a scientific intelligent transaction risk assessment system is of great significance for the healthy development of ecommerce platform.This article begins with the current status and problems of transaction risk assessment in china's ecommerce platform. Through research and analysis of transaction risk assessment methods in China and overseas with the guideline of risk management theory and the characteristics analysis of risk transaction, an alternative transaction risk index system is built and finally a transaction risk assessment system based on big data and data mining methods and social relation network analysis technique is established. It's main contributions are:Established an alternative ecommerce transaction risk index system; Trained a single customer risk assessment model based on gradient boosting decision tree algorithms; Optimized the single customer risk assessment model with logistic regression and credit score card technology; Established spammer community recognition model based on social network analysis; Assembled single customer risk assessment model and the spammer community recognition model to find more risk customer; Assessed the models in J company and proved better prediction results; Discussed transaction risk management and transaction risk legal issues.The contributions of this paper are following. First, based on the traditional risk management theory framework, innovative thinking and attempts are made at the technical level of transaction risk assessment and a comprehensive review of the transaction risk index system for the whole industry was conducted. Then, the latest technology such as big data and data mining and social networks to perform risk assessment research combined with traditional mature credit score card technology and integrate kinds of technology were applied. A deep research was conducted starting from the risk assessment for a single customer to the relations analysis of spammer community network, and finally to the establishment of a set of systematic and scientific risk assessment model. At last, this paper tries to apply the modularity algorithm in community detecting while the research of community detecting in china are mainly focused on network clustering analysis now, It achieved good results. The attempts in the paper has taken a big step for the methodology exploration of ecommerce transaction risk management. With the application of big data technology, we can access customer information with larger volume and velocity. Data mining method provides the risk recognition with more intelligence and more accuracy. Thus, user's abnormal behavior detection is more reliable. The exploration of these new technology has brought new ideas and means to the research of transaction risk management theory, as well as to the business practice in the field of ecommerce risk management.