System and method for confidence based incremental access authentication

• US 7,039,951 B1
• Filed: 06/06/2000
• Issued: 05/02/2006
• Est. Priority Date: 06/06/2000
• Status: Expired due to Term

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Ramaswamy, Ganesh N., Chaudhari, Upendra V.
• Primary Examiner(s): REVAK, CHRISTOPHER A
• Application Number:
  US09/588,521
• Time in Patent Office: 2,156 Days
• Field of Search: 713182-186, 713200-202, 713/155, 704/246, 379/88.02, 340/5.1, 340/5.82, 340/5.84, 340/5.52, 382/115
• US Class Current: 726/7
• CPC Class Codes: G10L 17/06 Decision making techniques; Pattern matching strategies

Abstract

A system and method for providing continuous confidence-based authentication. The present invention may be implemented in an incremental access authentication system for controlling access to secured data having various levels of security. During the course of a conversational session between user and machine, a confidence-based authentication system according to the present invention will periodically analyze the input speech of a user interacting with the system to compute a “confidence measure” for the validity of an original identity claim i provided by the user at the commencement of the dialog session. The “confidence measure” computation process according to the present invention is seamlessly integrated into the incremental access authentication system so that the system can tailor its interaction with the user based on its confidence in the original identity claim.

Claims

1. A computer implemented method for authenticating a user in a conversational system, comprising the steps of:

receiving an identity claim from a user upon commencement of a dialog session;

computing a confidence score using speech input from the user, wherein the confidence score is a measure of confidence in the validity of the identity claim; and

providing the user access to secured data having varying levels of security, wherein providing access comprises determining a level of secured data that may be accessed by the user based on the computed confidence score,

wherein computing a confidence score comprises computing a confidence score upon occurrence of a predetermined event during the dialog session with the user to determine a current measure of confidence in the validity of the identity claim, and changing a level of secured data that may be accessed by the user during the dialog session, if necessary, according to the current measure of confidence in the validity of the identity claim.

2. The computer implemented method of claim 1, further comprising the step of maintaining the confidence score as part of the system state.

3. The computer implemented method of claim 1, further comprising the steps of:

partitioning the secured data into a plurality of data classes;

assigning a security level to each of the data classes; and

constructing an access map based on the security levels for accessing the secured data.

4. The computer implemented method of claim 3, further comprising the steps of:

selecting a range of confidence scores;

partitioning the range of confidence scores into a plurality of regions; and

assigning each region to one of the security levels.

5. The computer implemented method of claim 4, wherein the step of providing the user access to secured data based on the computed confidence score comprises the steps of:

determining a given region of the plurality of regions which comprises the computed confidence score;

determining the security level assigned to the given region; and

accessing secured data using the access map based on the security level assigned to the given region.

6. The computer implemented method of claim 5, wherein the step of accessing secured data using the access map comprises the step of allowing access to secured data that is assigned to the security level of the given region and secured data assigned to at least one security level that is lower than the security level of the given region.

7. The computer implemented method of claim 1, wherein the predetermined event comprises expiration of a predetermined time period during the dialog session.

8. The computer implemented method of claim 1, wherein the predetermined event is a user query for accessing secured data.

9. The computer implemented method of claim 1, wherein the confidence score is based on a linear function of statistical models that characterize the score under a plurality of conditions.

10. The computer implemented method of claim 9, wherein the confidence score comprises one of (1) a first component for considering a single mode implementation and (2) the first component and a second component for considering a multi-modal implementation.

11. The computer implemented method of claim 10, wherein the confidence score comprises a mixing factor for weighting the first and second component in a multi-modal implementation.

12. A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform method steps for authenticating a user in a conversational system, the method comprising the steps of:

receiving an identity claim from a user upon commencement of a dialog session;

computing a confidence score using speech input from the user, wherein the confidence score is a measure of confidence in the validity of the identity claim; and

providing the user access to secured data having varying levels of security, wherein providing access comprises determining a level of secured data that may be accessed by the user based on the computed confidence score,

wherein computing a confidence score comprises computing a confidence score upon occurrence of a predetermined event during the dialog session with the user to determine a current measure of confidence in the validity of the identity claim, and changing a level of secured data that may be accessed by the user during the dialog session, if necessary, according to the current measure of confidence in the validity of the identity claim.

13. The program storage device of claim 12, further comprising instructions for performing the step of maintaining the confidence score as part of the system state.

14. The program storage device of claim 12, further comprising instructions for performing the steps of:

partitioning the secured data into a plurality of data classes;

assigning a security level to each of the data classes; and

constructing an access map based on the security levels for accessing the secured data.

15. The program storage device of claim 14, further comprising instructions for performing the steps of:

selecting a range of confidence scores;

partitioning the range of confidence scores into a plurality of regions; and

assigning each region to one of the security levels.

16. The program storage device of claim 15, wherein the instructions for performing the step of providing the user access to secured data based on the computed confidence score comprise instructions for performing the steps of:

determining a given region of the plurality of regions which comprises the computed confidence score;

determining the security level assigned to the given region; and

accessing secured data using the access map based on the security level assigned to the given region.

17. The program storage device of claim 16, wherein the instructions for performing the step of accessing secured data using the access map comprise instructions for performing the step of allowing access to secured data that is assigned to the security level of the given region and secured data assigned to at least one security level that is lower than the security level of the given region.

18. The program storage device of claim 12, wherein the predetermined event comprises expiration of a predetermined time period during the dialog session.

19. The program storage device of claim 12, wherein the predetermined event is a user query for accessing secured data.

20. The program storage device of claim 12, wherein the confidence score is based on a linear function of statistical models that characterize the score under a plurality of conditions.

21. The program storage device of claim 20, wherein the confidence score comprises one of (1) a first component for considering a single mode implementation and (2) the first component and a second component for considering multi-modal implementation.

22. The program storage device of claim 21, wherein the confidence score comprises a mixing factor for weighting the first and second component in a multi-modal implementation.

23. A computer system to control incremental access authentication to secured data, the computer systems comprising:

a database that is partitioned into a plurality of data classes, wherein each data class is assigned a range of confidence scores based on a security level of the data class;

a computation module that computes a confidence score using speech input from a user, wherein the confidence score is a measure of confidence in the validity of an identity claim input by the user upon commencement of a dialog session, wherein the computation module computes a confidence score upon occurrence of a predetermined event during the dialog session with the user to determine a current measure of confidence in the validity of the identity claim; and

a dialog manager for controlling access to data in the database during the dialog session, wherein a level of access to the data classes is limited to a data class in which a current confidence score meets or exceeds a confidence score assigned to that data class, and wherein the level of access to the data classes can be changed during the dialog session, if necessary, according to the current measure of confidence in the validity of the identity claim.

24. The system of claim 23, further comprising an access map for mapping each data class with the corresponding range of confidence scores, wherein the access map is utilized by the dialog manager to provide access to data based on the current confidence score.

25. The system of claim 23, further comprising means for maintaining the current confidence score as part of the system state.

26. The system of claim 23, wherein the confidence score is based on a linear function of statistical models that characterize the score under a plurality of conditions.

27. The system of claim 26, wherein the confidence score comprises one of (1) a first component for considering a single mode implementation and (2) the first component and a second component for considering a multi-modal implementation.

28. The system of claim 27, wherein the confidence score comprises a mixing factor for weighting the first and second component in a multi-modal implementation.