System and method for identity-based fraud detection
View Patent ↗A method for identifying a fraudulent account application includes receiving a new account application comprising a plurality of identity-related fields and linking the identity-related fields associated with the new account application with identity-related fields associated with a plurality of historical account applications. The links form a graphical pattern on which statistical analysis can performed to determine the likelihood that the new account application is fraudulent. The statistical analysis can comprise comparing the graphical pattern to a known, or normal graphical pattern in order to detect differences, or anomalies occurring in the graphical pattern associated with the new account application.
1. A method for determining a likelihood of fraud associated with an input identity record, comprising:
receiving an identity record comprising identity related information, wherein the received identity record comprises fields configured to store the identity related information, and wherein the fields form nodes;
retrieving identity records from a database of identity records generated from a plurality of identity records received from a plurality of client institutions;
linking the received identity record with the retrieved identity records based on the identity related information, wherein the plurality of linked identity records are stored as a graphical network, and wherein the graphical network comprises a plurality of graphs defined by sets of nodes and sets of edges composed of pairs of nodes sharing a common characteristic;
evaluating the received identity record using the linked, retrieved identity records;
generating a fraud score indicative of the likelihood of fraud associated with the input identity record based on the evaluation of the received identity record;
returning the fraud score to a client institution;
receiving updated identity related information from the client institution based on challenge response interrogations initiated in response to the generated fraud score; and
providing enhanced indication of the likelihood of fraud associated with the input identity record.
2. The method of claim 1 , wherein the plurality of graphs are combined into a single graph.
3. The method of claim 1 , further comprising linking nodes where the characteristic associated with the nodes are identical.
4. The method of claim 1 , further comprising linking nodes where the characteristic associated with the nodes are substantially similar.
5. The method of claim 1 , wherein the nodes are given fuzzy values.
6. The method of claim 5 , further comprising linking nodes based on a comparison of the fuzzy values associated with the nodes.
7. The method of claim 1 , wherein the graphical network is stored as an adjacency matrix.
8. The method of claim 7 , wherein only the non-zero elements of the adjacency matrix are stored.
9. The method of claim 1 , further comprising storing auxiliary information with at least some of the nodes in the graphical network.
10. The method of claim 1 , further comprising storing auxiliary information with at least some of the edges comprising the graphical network.