System and method for the detection and termination of fraudulent services

• US 7,512,221 B2
• Filed: 01/28/2005
• Issued: 03/31/2009
• Est. Priority Date: 07/31/2002
• Status: Expired due to Fees

• Current Assignee: Cerebrus Solutions Ltd (Neural Technologies Limited)
• Original Assignee: Cerebrus Solutions Ltd (Neural Technologies Limited)
• Inventors: Toms, Alvin David
• Primary Examiner(s): TRAN, QUOC DUC
• Application Number:
  US11/046,200
• Publication Number:
  US 20050185779A1
• Time in Patent Office: 1,523 Days
• Field of Search: 379/114.14, 379/127.02, 379/144.03, 379/188, 379/189, 379/195, 379/196, 379/88.01, 379/88.02, 455/410, 455/411, 704/221, 704/243, 704/246
• US Class Current: 379/114.14
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• CPC Class Codes: G10L 17/22 Interactive procedures; Man-machine interfaces
• CPC Class Codes: H04M 1/67 by electronic means
• CPC Class Codes: H04M 2201/18 Comparators
• CPC Class Codes: H04M 2201/40 using speech recognition
• CPC Class Codes: H04M 2201/41 using speaker recognition
• CPC Class Codes: H04M 2203/6027 Fraud preventions
• CPC Class Codes: H04M 2207/206 composed of PSTN and wireless network
• CPC Class Codes: H04M 3/2218 Call detail recording
• CPC Class Codes: H04M 3/2281 Call monitoring, e.g. for law enforcement purposes; Call tracing; Detection or prevention of malicious calls
• CPC Class Codes: H04M 3/36 Statistical metering, e.g. recording occasions when traffic exceeds capacity of trunks
• CPC Class Codes: H04M 3/42059 Making use of the calling party identifier

Abstract

A system is disclosed for the automatic detection of fraudulent activity on a transaction network, for which each transaction over the network has an associated identifier. In one embodiment, the system includes voice comparison means for comparing a first sampled voice of a user of a first transaction with a subsequently sampled voice of a user of a subsequent transaction having an identical identifier to that of the first transaction. Control means in the form of a voice-based fraud detection engine is provided for determining, from said comparison, a profile of user usage that is representative of a total number of different users of the associated identifier. In a preferred embodiment, the system also includes voice sampling means for sampling a voice of the user of the first transaction to generate a first voice sample.

Claims

1. A system for the automatic detection of fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the system comprising:

voice comparison means for comparing a first sampled voice of a user of a first transaction with a subsequently sampled voice of a user of a subsequent transaction having an identical identifier to that of the first transaction, and control means for determining, from said comparison, a profile of user usage that is representative of a number of different users of the identifier, for comparing the profile of user usage with a threshold for fraudulent use and for generating a fraud condition signal in the event that the profile of user usage exceeds said threshold.

2. The system as claimed in claim 1, further comprising voice sampling means for sampling a voice of the user of the first transaction to generate a first voice sample, and for sampling a voice of the user of the subsequent transaction to generate a subsequent voice sample.

3. The system as claimed in claim 2, wherein said voice comparison means is arranged to generate a first voice print from said first voice sample.

4. The system as claimed in claim 3, further comprising voice print storage means for storing said first voice print.

5. The system as claimed in claim 3, further comprising selection means configured to initiate voice sampling by the voice sampling means only for selected transactions.

6. The system as claimed in claim 5, wherein the selection means is configured to periodically initiate voice sampling of transactions.

7. The system as claimed in claim 5, wherein the selection means includes a fraud detection system for determining whether a transaction has a fraudulent characteristic and for generating a fraud alert signal in the event thereof, and whereby voice sampling is initiated in response to generation of the fraud alert signal.

8. The system as claimed in claim 7, wherein the fraud detection system includes means for comparing a use profile for the identifier with one or more characteristics of a transaction associated with the identifier, thereby to determine whether the transaction has a fraudulent characteristic.

9. The system as claimed in claim 3, wherein the voice sampling means is configured to sample a voice of a user several times for each transaction that is selected for voice sampling.

10. The system as claimed in claims 3, wherein the voice comparison means is arranged to generate a positive comparative output signal for each subsequent voice sample which is substantially matched with a stored voice print and a negative comparative output signal for each subsequent voice sample which does not substantially match a stored voice print.

11. The system as claimed in claim 10, wherein the profile of user usage for an identifier includes a representation of a total number of negative comparative output signals for said identifier, and wherein said threshold for fraudulent use is a maximum number of different users of the identifier above which there is an indication of fraudulent use.

12. The system as claimed in claim 10, wherein the profile of user usage includes a representation of a percentage of the total number of comparative output signals that are negative comparative output signals.

13. The system as claimed in claim 10, wherein the control means includes halt signal means for generating a halt signal which serves to halt analysis of transactions having an identifier for which the profile of user usage provides a positive indication of non-fraudulent use.

14. The system as claimed in claim 13, wherein the control means is arranged to compare the profile of user usage with a threshold for non-fraudulent use, and wherein the halt signal means is arranged to generate the halt signal in the event that the profile of user usage exceeds said threshold for non-fraudulent use.

15. The system as claimed in claim 14, wherein the profile of user usage includes a representation of a total number of positive comparative output signals for said identifier, and wherein said threshold for non-fraudulent use is a minimum number of transactions having the same user which is indicative of non-fraudulent use.

16. The system as claimed in claim 14, wherein the profile of user usage includes a representation of a percentage of the total number of comparative output signals that are positive comparative output signals.

17. The system as claimed in claim 3, wherein the profile of user usage includes a representation of one or more characteristics of the associated identifier.

18. The system as claimed in claim 17, wherein the control means includes adjustment means for adjusting at least one of the threshold for fraudulent use and the threshold for non-fraudulent use for an associated identifier in response to one or more of said associated identifier characteristics.

19. The system as claimed in claim 1, further comprising a record storage means for storing data records that are representative of the profile of user usages.

20. The system as claimed in claim 1, wherein the identifier includes data of alphabetic, numeric or alphanumeric character.

21. The system as claimed in claim 1, wherein the transaction network is a telecommunications network.

22. The system as claimed in claim 21, wherein the transaction is a telephone call.

23. The system as claimed in claim 22, wherein the identifier is a telephone number unique to a mobile telephone or to a landline telephone connection.

24. The system as claimed in claim 21, wherein the transaction is a payment for goods and/or services.

25. The system as claimed in claim 24, wherein the identifier is a payment card number.

26. The system as claimed in claim 1, wherein the control means is arranged to automatically supply the fraud condition signal to the transaction network, thereby to implement automatic termination of a fraudulent transaction.

27. A system for the automatic detection of fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the system comprising a voice comparator for comparing a first sampled voice of a user of a first transaction with a second sampled voice of a user of a subsequent transaction having an identical identifier to that of the first transaction, and a controller for determining, from said comparison, a profile of user usage for the identifier that is representative of how many different users of the identifier there are, for comparing the profile of user usage with a threshold for fraudulent use and for generating a fraud condition signal in the event that the profile of user usage exceeds said threshold.

28. A method of automatic detection of fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the method comprising:

comparing a first sampled voice of a user of a first transaction with a subsequent sampled voice of a user of a subsequent transaction having an identical identifier to that of the first transaction, determining, from said comparison, a profile of user usage for said identifier that is representative of a number of different users of the identifier, comparing the profile of user usage with a threshold for fraudulent use, and generating a fraud condition signal in the event that the profile of user usage exceeds said threshold.

29. A method as claimed in claim 28, including sampling a voice of the user of the first transaction to generate a first voice sample, generating a first voice print from said first voice sample and storing the first voice print with the associated identifier in a voice print archive, sampling a voice of the user of the subsequent transaction to generate a subsequent voice sample, and comparing the subsequent voice sample with the first voice print for the purpose of determining the profile of user usage for said identifier.

30. The method as claimed in claim 29, including automatically supplying the fraud condition signal to the transaction system to implement automatic termination of a fraudulent transaction.

31. The method as claimed in claim 29 additionally comprising:

providing an initial indication of whether a transaction has a fraudulent characteristic using fraud detection means,

generating a fraud alert signal in the event that the initial indication is that the transaction has a fraudulent characteristic, and

selectively initiating voice sampling in response to generation of the fraud alert signal by the fraud detection system.

32. A fraud detection system for detecting fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the system comprising:

a voice comparator for comparing a first sampled voice of a user of one of the identifiers with a subsequently sampled voice of a user of the same identifier taken during a subsequent use of said same identifier;

a profiler for building a profile of usage by users of said same identifier based on the voice comparisons, such that the profile is representative of a number of different users of the identifier; and

,

a fraud detector for comparing the number of different users represented in the profile as using the identifier to a threshold indicative of fraudulent use, wherein in the event that the number of users exceeds the threshold a fraud alert signal is generated.

33. A method of automatic detection of fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the method comprising:

sampling a first voice of a user of one of the identifiers;

subsequently sampling a second voice of a user of the same identifier;

comparing the first sampled voice with the subsequently sampled voice;

profiling usage by users of said same identifier based on the voice comparisons, such that the profile is representative of a number of different users of the identifier;

comparing the number of different users represented in the profile as using the identifier to a threshold indicative of fraudulent use; and

,

generating a fraud alert signal in the event that the number of users exceeds said threshold.

34. A fraud detection system for detecting fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the system comprising:

a voice based fraud detection engine for sampling the voice of each user of the identifier during each use of the identifier using a voice analysis technique to identify the number of unique users of the identifier and to indicate the presence of fraudulent use when the number of unique users of the identifier exceeds a threshold; and

,

a fraud alert generator to signal presence of fraudulent activity when the voice based fraud detection engine indicates the presence of fraud.

37. A fraud detection system as claimed in claim 34, wherein the voice based fraud detection engine of the system comprises:

a voice comparator for comparing a first sampled voice of a user of one of the identifiers with a subsequently sampled voice of a user of the same identifier taken during a subsequent use of said same identifier;

a profiler for building a profile of usage by users of said same identifier based on the voice comparisons, such that the profile is representative of a number of different users of the identifier; and

,

a fraud detector for comparing the number of different users represented in the profile as using the identifier to a threshold indicative of fraudulent use, wherein in the event that the number of users exceeds the threshold a fraud alert signal is generated.

35. A fraud detection system for detecting fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the system comprising:

a use characteristics fraud detection engine for analysing characteristics of the use of the network to determine whether fraud is indicated in the use of one of the identifiers;

a voice based fraud detection engine for sampling the voice of each user of the identifier during each use of the identifier using a voice analysis technique to identify the number of unique users of the identifier and to indicate the presence of fraudulent use when the number of unique users of the identifier exceeds a threshold; and

,

a fraud alert generator to signal presence of fraudulent activity when both of the use characteristics fraud detection engine and the voice based fraud detection engine indicate the presence of fraud.

36. A fraud detection system as claimed in claim 35, wherein use characteristics fraud detection engine analyses one or more of the following to determine that there is an indication of fraudulent use of the identifier:

the rate of use of the identifier, the manner of use of the identifier, the duration of each use of the identifier, or the types of transaction conducted with the identifier.

38. A method of automatic detection of fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the method comprising:

sampling the voice of each user during each use of said identifier and using a voice analysis technique to identify the number of unique users of the identifier to indicate the presence of fraudulent use in the event that the number of unique users of the identifier exceeds a threshold indicative of fraudulent use; and

,

generating a fraud alert for signalling presence of fraudulent activity when the voice based analysis indicates the presence of fraud.

41. A method as claimed in claim 38, wherein the voice analysis technique of the method comprises the steps of:

comparing one of the sampled voices with a subsequently sampled other of the voices;

profiling usage by users of said same identifier based on the voice comparisons, such that the profile is representative of a number of different users of the identifier;

comparing the number of different users represented in the profile as using the identifier to a threshold indicative of fraudulent use; and

,

generating a fraud alert signal in the event that the number of users exceeds said threshold.

39. A method of automatic detection of fraudulent activity on a transaction network, for which each transaction over the network has an identifier associated therewith, the method comprising:

analysing characteristics of the use of the network to determine whether fraud is indicated in the use of one of the identifiers;

sampling the voice of each user during each use of said identifier and using a voice analysis technique to identify the number of unique users of the identifier to indicate the presence of fraudulent use in the event that the number of unique users of the identifier exceeds a threshold indicative of fraudulent use; and

,

generating a fraud alert for signalling presence of fraudulent activity when both of the use characteristics analysis and the voice based analysis indicate the presence of fraud.

40. A method as claimed in claim 39, wherein the analysis of characteristics of the use of the network includes the steps of:

analysing one or more of the following to determine that there is an indication of fraudulent use of the identifier;

the rate of use of the identifier, the manner of use of the identifier, the duration of each use of the identifier, or the types of transaction conducted with the identifier.