Method and apparatus for privacy preserving data mining by restricting attribute choice

• US 20070233711A1
• Filed: 04/04/2006
• Published: 10/04/2007
• Est. Priority Date: 04/04/2006
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Halim, Nagui, Aggarwal, Charu
• Granted Patent: US 7,475,085 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06F 21/6254 by anonymising data, e.g. decorrelating personal data from the owner's identification
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99939 Privileged access
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99944 Object-oriented database structure

Abstract

Improved techniques for privacy preserving data mining of multidimensional data records are disclosed. For example, a technique for generating at least one output data set from at least one input data set for use in association with a data mining process comprises the following steps/operations. At least one relevant attribute of the at least one input data set is selected through determination of at least one relevance coefficient. The at least one output data set is generated from the at least one input data set, wherein the at least one output data set comprises the at least one relevant attribute of the at least one input data set, as determined by use of the at least one relevance coefficient.

Claims

1. A method of generating at least one output data set from at least one input data set for use in association with a data mining process, comprising the steps of:

selecting at least one relevant attribute of the at least one input data set through determination of at least one relevance coefficient; and

generating the at least one output data set from the at least one input data set wherein the at least one output data set comprises the at least one relevant attribute of the at least one input data set, as determined by use of the at least one relevance coefficient.

2. The method of claim 1, wherein the output data set is anonymized.

3. The method of claim 2, wherein the anonymization generates synthetic pseudo-data.

4. The method of claim 1, wherein the data mining process is a privacy preserving data mining process.

5. The method of claim 1, wherein the at least one input data set is multidimensional.

6. The method of claim 1, wherein the at least one relevance coefficient is computed using a quantitative measure of the effect of its deletion on the data mining process.

7. The method of claim 1, wherein a global feature analysis may be used for computing the at least one relevance coefficient.

8. The method of claim 1, wherein a local feature analysis may be used for computing the at least one relevance coefficient.

9. The method of claim 1, wherein the at least one relevance coefficient may be used in at least one distance function during a condensation process.

10. The method of claim 7, wherein at least one distance function is more strongly weighted by features which have at least one high relevance coefficient.

11. The method of claim 1, wherein the features with one or more relevance coefficients are pre-selected for privacy preservation purposes.

12. Apparatus for generating at least one output data set from at least one input data set for use in association with a data mining process, the apparatus comprising:

a memory; and

at least one processor coupled to the memory and operative to;

(i) select at least one relevant attribute of the at least one input data set through determination of at least one relevance coefficient and (ii) generate the at least one output data set from the at least one input data set wherein the output data set comprises the at least relevant attribute of the at least one input data set, as determined by use of the at least one relevance coefficient.

13. The apparatus of claim 12, wherein the output data set is anonymized.

14. The apparatus of claim 13, wherein the anonymization generates synthetic pseudo-data.

15. The apparatus of claim 12, wherein the data mining process is a privacy preserving data mining process.

16. The apparatus of claim 12, wherein the at least one input data set is multidimensional.

17. The apparatus of claim 12, wherein the at least one relevance coefficient is computed using a quantitative measure of the effect of its deletion on the data mining application.

18. The apparatus of claim 12, wherein a global feature analysis may be used for computing the at least one relevance coefficient.

19. The apparatus of claim 12, wherein a local feature analysis may be used for computing the at least one relevance coefficient.

20. The apparatus of claim 12, wherein the at least one relevance coefficient may be used in at least one distance function during a condensation process.

22. The apparatus of claim 12, wherein the features with at least one high relevance coefficient are pre-selected for privacy preservation purposes.

21. The method of claim 21, wherein the at least one distance function is more strongly weighted by features which have at least one relevance coefficient.

23. An article of manufacture for generating at least one output data set from at least one input data set for use in association with a data mining process, comprising a machine readable medium containing one or more programs which when executed implement the steps of:

selecting at least one relevant attribute of the at least one input data set through determination of at least one relevance coefficient and generating the at least one output data set from the at least one input data set

wherein the at least one output data set comprises only the most relevant attributes of the at least one input data set, as determined by use of the at least one relevance coefficient.