Data anonymization based on guessing anonymity

• US 10,380,351 B2
• Filed: 01/06/2014
• Issued: 08/13/2019
• Est. Priority Date: 12/18/2008
• Status: Active Grant

• Current Assignee: Accenture Global Services Limited (Accenture PLC)
• Original Assignee: Accenture Global Services Limited (Accenture PLC)
• Inventors: Rachlin, Yaron, Probst, Katharina, Ghani, Rayid
• Primary Examiner(s): Obeid, Mamon
• Application Number:
  US14/148,237
• Publication Number:
  US 20140123304A1
• Time in Patent Office: 2,045 Days
• Field of Search: 705 26, 705 30, 705 50, 705 51
• CPC Class Codes: G06F 21/60 Protecting data
• CPC Class Codes: G06F 21/6254 by anonymising data, e.g. decorrelating personal data from the owner's identification
• CPC Class Codes: H04L 9/30 Public key, i.e. encryption algorithm being computationally infeasible to invert or user's encryption keys not requiring secrecy

Abstract

Privacy is defined in the context of a guessing game based on the so-called guessing inequality. The privacy of a sanitized record, i.e., guessing anonymity, is defined by the number of guesses an attacker needs to correctly guess an original record used to generate a sanitized record. Using this definition, optimization problems are formulated that optimize a second anonymization parameter (privacy or data distortion) given constraints on a first anonymization parameter (data distortion or privacy, respectively). Optimization is performed across a spectrum of possible values for at least one noise parameter within a noise model. Noise is then generated based on the noise parameter value(s) and applied to the data, which may comprise real and/or categorical data. Prior to anonymization, the data may have identifiers suppressed, whereas outlier data values in the noise perturbed data may be likewise modified to further ensure privacy.

Claims

1. A method comprising:

invoking, by a processing device, an anonymization function provided via an application interface,

the anonymization function being invoked through an activation of a user input mechanism of the application interface;

receiving, by the processing device and based on invoking the anonymization function, a selection of a noise model to be applied to data;

determining, by the processing device, a first anonymization parameter,

the first anonymization parameter comprising an expected distortion;

determining, by the processing device, a particular level for the first anonymization parameter,

the determined particular level being a maximum expected distortion level;

determining, by the processing device, a second anonymization parameter based on the first anonymization parameter,

the second anonymization parameter being an expected guessing anonymity,

the expected guessing anonymity being a number of guesses needed to determine an original record from a sanitized record;

determining, by the processing device and based on determining the particular level for the first anonymization parameter, a parameter value for the noise model that optimizes the second anonymization parameter;

generating, by the processing device, noise based on the noise model and the determined parameter value;

applying, by the processing device, the generated noise to the data;

generating, by the processing device, noise perturbed data based on applying the generated noise to the data,

the noise perturbed data being associated with;

the determined particular level with respect to the expected distortion, and

the determined parameter value with respect to the expected guessing anonymity; and

providing, by the processing device, the noise perturbed data to a destination.

2. The method of claim 1, further comprising:

modifying outlier data in the noise perturbed data.

3. The method of claim 1, where applying the generated noise comprises:

applying the generated noise to at least one identifier associated with the data.

4. The method of claim 1, further comprising:

modifying data in the noise perturbed data,

where providing the noise perturbed data comprises;

providing the noise perturbed data with the modified data.

5. The method of claim 1, where optimizing the second anonymization parameter comprises:

maximizing the expected guessing anonymity based on the maximum expected distortion.

18. The method of claim 1, further comprising:

receiving a selection of one or more attributes of the data; and

where applying the generated noise to the data comprises;

applying the generated noise to the one or more attributes of the data.

6. An apparatus comprising:

a memory to store instructions; and

a processor to execute the instructions to;

invoke an anonymization function provided via an application interface,

the anonymization function being invoked through an activation of a user input mechanism of the application interface;

receive, based on invoking the anonymization function, a selection of a noise model to be applied to data;

determine a first anonymization parameter,

the first anonymization parameter comprising an expected distortion;

determine a particular level for the first anonymization parameter;

the determined particular level being a maximum expected distortion level;

determine a second anonymization parameter based on the first anonymization parameter,

the second anonymization parameter being an expected guessing anonymity,

the expected guessing anonymity being a number of guesses needed to determine an original record from a sanitized record;

determine, based on determining the particular level for the first anonymization parameter, a parameter value for the noise model that optimizes the second anonymization parameter;

generate noise based on the noise model and the determined parameter value;

apply the generated noise to the data;

generate noise perturbed data based on applying the generated noise to the data,

the noise perturbed data having an anonymized structure, and

the noise perturbed data being associated with;

the determined particular level with respect to the expected distortion, and

the determined parameter value with respect to the expected guessing anonymity; and

provide the noise perturbed data to a destination.

7. The apparatus of claim 6, where the processor, when optimizing the second anonymization parameter, is to:

optimize the second anonymization parameter by maximizing the expected guessing anonymity based on the maximum expected distortion level.

8. The apparatus of claim 6, where the processor, when applying the generated noise, is to:

apply the generated noise to at least one quasi-identifier in the data.

9. The apparatus of claim 6, where the processor is further to:

modify outlier data in the noise perturbed data.

10. The apparatus of claim 6, where the processor is further to:

suppress one or more identifiers in the data.

11. The apparatus of claim 6, where the noise model is applicable to:

real attributes in the data, and

categorical attributes in the data.

17. The apparatus of claim 6, where a type of data is specified based on the selection of the noise model.

19. The apparatus of claim 6, where the processor is further to:

receive a selection of one or more attributes of the data; and

where the processors, when applying the generated noise to the data, are to;

apply the generated noise to the one or more attributes of the data.

12. A non-transitory computer readable medium storing instructions, the instructions comprising:

one or more instructions which, when executed by a processor, cause the processor to;

invoke an anonymization function provided via an application interface,

the anonymization function being invoked through an activation of a user input mechanism of the application interface;

receive, based on invoking the anonymization function a selection of a noise model to be applied to data;

determine a first anonymization parameter;

the first anonymization parameter comprising an expected distortion;

determine a particular level for the first anonymization parameter;

the determined particular level being a maximum expected distortion level;

determine a second anonymization parameter based on the first anonymization parameter,

the second anonymization parameter being an expected guessing anonymity,

the expected guessing anonymity being a number of guesses needed to determine an original record from a sanitized record;

determine, based on determining the particular level for the first anonymization parameter, a parameter value for the noise model that optimizes the second anonymization parameter;

generate noise based on the noise model and the determined parameter value;

apply the generated noise to the data;

generate noise perturbed data based on applying the generated noise to the data,

the noise perturbed data being associated with;

the determined particular level with respect to the expected distortion, and

the determined parameter value with respect to the expected guessing anonymity; and

provide the noise perturbed data to a destination.

13. The non-transitory computer readable medium of claim 12, where the one or more instructions, when executed by the processor, further cause the processor to:

modify outlier data in the noise perturbed data.

14. The non-transitory computer readable medium of claim 12, where the one or more instructions, that cause the processor to apply the generated noise to the data, cause the processor to:

apply the generated noise to at least one identifier associated with the data.

15. The non-transitory computer readable medium of claim 12, where the one or more instructions, when executed by the processor, further cause the processor to:

modify the noise perturbed data, and

where the one or more instructions, that cause the processor provide the noise perturbed data, cause the processor to;

provide the modified noise perturbed data based on modifying the noise perturbed data.

16. The non-transitory computer readable medium of claim 12, where the one or more instructions, that cause the processor to optimize the second anonymization parameter, cause the processor to:

maximize the expected guessing anonymity subject to the maximum expected distortion level.

20. The non-transitory computer readable medium of claim 12, where the one or more instructions, that cause the processor to optimize the second anonymization parameter, cause the processor to:

receive a selection of one or more attributes of the data; and

where the one or more instructions, that cause the processor to apply the generated noise to the data, cause the processor to;

apply the generated noise to the one or more attributes of the data.