Research On Unsupervised Link Discovery Technology In Exceptional Exchange Investigation

Financial crimes always are associated with fund flow of account, which is more serious in money-laundering. At present, most systems take use of traditional knowledge discovery. Firstly, we propose interested suppose by investigating data set; and then, analysis approach will be designed to solve these problems. However, the approach always has a disadvantage that we are difficult of discriminating those unknown knowledge models. For the problem, we make use of constraint-based unsupervised link discovery finding the user-interested information in separation phase.According to social network analysis theory, a large number of abnormal behaviors are always hidden in the normal mode of behavior. Corresponding to the field of financial supervising is to remove the normal information of the transaction, which highlights unusual information. It's a proceeding of data mining among account exchanges. The goal of constraint-based unsupervised link discovery, which is the development of link analysis in data mining, is to build a network using many irrelevant objects, and to find out some useful modes and patterns. The flow includes three phases as follow: phase of designing exchange mode, phase of detecting abnormal exchange path type, phase of detecting abnormal exchange node.Designing exchange mode is a pretreatment phase, which aims at transferring the exchange data to mode graph. The key in this phase is to establish object type and object relationship. According to anti-money-laundering knowledge and experts' summarization, the question for discussion focuses five node object type as follows: accounts, money, organization, trade time, trade address. So the exchange network mode is composed of these five node type and relationship each other.Detecting abnormal exchange path type, which is based of Graph Unusual Path Algorithm in an unsupervised link discovery method, makes use of "rarity" quantifying the user's interest. By detecting abnormal exchange path type, we can find out some unconspicuous unusual information. Detecting abnormal exchange node, which is based of abnormal exchange path type discovery, makes use of "frequency" quantifying the user's interest. By detecting abnormal exchange node, we can find out some unusual information which always interest users.