频繁子图挖掘算法及其在洗钱模式发现中的应用分析-frequent subgraph mining algorithm and its application analysis in money laundering pattern discovery.docxVIP

AbstractAs the transaction between finance account has an in-out relationship naturally,a group that have trade relationship must form a network by the contact of accounts,and can be expressed by graph type directly.According to the finance anti-laundering research,there are lots of trade models in money-laundering behaviors,which corresponding to the subgraph models in trade network.Using frequent subgraph mining methods can help finding the frequent ones in these models.Analysis shows that the financial transaction network can use a map to the directed,weighted model. With nodes that account, while that transaction, the direction that the transfer of funds between accounts direction, the weight based on transaction information to set. Consider the background of very complex financial transactions, select one of several key attributes such as the amount, type-related enterprises, trading areas statute, into the weight transactions.Nowadays,the developed frequent subgraph mining algorithms aim at non-direction,non-weight marked graph mining mostly,but few aim at directed graph pattern mining.The graph pattern means a uniform expresstion of graphs that have different marked nodes but same structure.Most frequent subgraph mining algorithms treat them different patterns.The mSpan frequent pattern mining algorithm base on FP-growth uses a min expand way gets a min edgess code and a abstract nodess codes sequence.It can mark a graph pattern one and only,and solve the graph patterns isomorphic problem and the redundant expand problem.The algorithm combins the code of the direction of an edge into the weight of the edge as the instance of the expand direction.The experiment shows mSpan can mine all frequent directed,weighted graph models.Keywords:financial transaction network,frequent subgraph mining,modeling目录摘要............................................................................................................ (1)ABSTRACT...........................................