Techniques for data classification based on sensitive data

1. A computer-implemented method for data loss prevention, the method being performed by a computing device comprising at least one processor, the method comprising:

selecting, by the at least one processor, from a group of electronic files within a file storage a sample set representing fewer than all of the electronic files in the file storage;

classifying, by the at least one processor, each electronic file in the sample set, by using machine learning to build statistical models using example documents and calculating content similarity, wherein each electronic file within the sample set found to contain sensitive information is classified as a sensitive file and each electronic file within the sample set found to lack sensitive information is classified as a nonsensitive file;

further classifying each sensitive file into a category based on a type of sensitive information contained in the sensitive file, wherein the type of sensitive information includes at least one of personally identifiable information, medical data, financial data, or source code;

generating, by the at least one processor, an estimate for the group of electronic files based on the classifications of each electronic file in the sample set, wherein the estimate identifies a volume of sensitive information within the group of electronic files and the category of sensitive information within the group of electronic files; and

applying, by the at least one processor, a safeguard to the group of electronic files when the estimate of the volume of sensitive information within the group of electronic files exceeds a predetermined quantity,

wherein a number of electronic files selected for the sample set is based on a desired level of accuracy for the estimate of sensitive information such that an error rate for the estimate of sensitive information is substantially constrained within a predetermined percentage range, and

wherein the safeguard flags for further investigation a user's access of an electronic file in the group of electronic files based on the category of sensitive information identified for the group of electronic files.

2. The method of claim 1 , wherein the desired level of accuracy is specified by a user who initiates a data scan of the group of electronic files.

3. The method of claim 1 , wherein:

the safeguard also flags for further investigation the user's access of an electronic file in the group of electronic files based on the estimate of the volume of sensitive information identified for the group of electronic files.

4. The method of claim 1 , wherein selecting the sample set from the group of electronic files comprises randomly selecting, by the at least one processor, electronic files from the group of electronic files until enough electronic files are selected to meet a predetermined sampling threshold.

5. At least one non-transitory processor readable storage medium storing a computer program of instructions configured to be readable by at least one processor for instructing the at least one processor to execute a computer process for performing the method as recited in claim 1 .

6. An article of manufacture for data loss prevention comprising:

at least one processor readable storage medium; and

instructions stored on the at least one medium;

wherein the instructions are configured to be readable from the at least one medium by at least one processor and thereby cause the at least one processor to operate so as to:

select from a group of electronic files within a file storage a sample set representing fewer than all of the electronic files in the file storage;

classify each electronic file in the sample set, by using machine learning to build statistical models using example documents and calculating content similarity, wherein each electronic file within the sample set found to contain sensitive information is classified as a sensitive file and each electronic file within the sample set found to lack sensitive information is classified as a nonsensitive file;

further classify each sensitive file into a category based on a type of sensitive information contained in the sensitive file, wherein the type of sensitive information includes at least one of personally identifiable information, medical data, financial data, or source code;

generate an estimate for the group of electronic files based on the classifications of each electronic file in the sample set, wherein the estimate identifies a volume of sensitive information within the group of electronic files and the category of sensitive information within the group of electronic files; and

apply a safeguard to the group of electronic files when the estimate of the volume of sensitive information within the group of electronic files exceeds a predetermined quantity,

wherein a number of electronic files selected for the sample set is based on a desired level of accuracy for the estimate of sensitive information such that an error rate for the estimate of sensitive information is substantially constrained within a predetermined percentage range, and

wherein the safeguard flags for further investigation a user's access of an electronic file in the group of electronic files based on the category of sensitive information identified for the group of electronic files.

7. The article of claim 6 , wherein the desired level of accuracy is specified by a user who initiates a data scan of the group of electronic files.

8. The article of claim 6 , wherein the safeguard also flags for further investigation the user's access of an electronic file in the group of electronic files based on the estimate of the volume of sensitive information identified for the group of electronic files.

9. The article of claim 6 , wherein selecting the sample set from the group of electronic files comprises randomly selecting electronic files from the group of electronic files until enough electronic files are selected to meet a predetermined sampling threshold.

10. A system for data loss prevention comprising:

one or more processors communicatively coupled to a network; wherein the one or more processors are configured to:

select from a group of electronic files within a file storage a sample set representing fewer than all of the electronic files in the file storage;

classify each electronic file in the sample set, by using machine learning to build statistical models using example documents and calculating content similarity, wherein each electronic file within the sample set found to contain sensitive information is classified as a sensitive file and each electronic file within the sample set found to lack sensitive information is classified as a nonsensitive file;

further classify each sensitive file into a category based on a type of sensitive information contained in the sensitive file, wherein the type of sensitive information includes at least one of personally identifiable information, medical data, financial data, or source code;

generate an estimate for the group of electronic files based on the classifications of each electronic file in the sample set, wherein the estimate identifies a volume of sensitive information within the group of electronic files and the category of sensitive information within the group of electronic files; and

apply a safeguard to the group of electronic files when the estimate of the volume of sensitive information within the group of electronic files exceeds a predetermined quantity,

wherein a number of electronic files selected for the sample set is based on a desired level of accuracy for the estimate of sensitive information such that an error rate for the estimate of sensitive information is substantially constrained within a predetermined percentage range, and

wherein the safeguard flags for further investigation a user's access of an electronic file in the group of electronic files based on the category of sensitive information identified for the group of electronic files.

11. The system of claim 10 , wherein the safeguard also flags for further investigation the user's access of an electronic file in the group of electronic files based on the estimate of the volume of sensitive information identified for the group of electronic files.

12. The system of claim 10 , wherein selecting the sample set from the group of electronic files comprises randomly selecting electronic files from the group of electronic files until enough electronic files are selected to meet a predetermined sampling threshold.

13. The method of claim 1 further comprising, prior to the step of selecting the sample set of files:

identifying the total number of electronic files in the file storage; and

determining that the total number of electronic files exceeds a threshold amount.

14. The article of claim 6 , wherein the instructions are configured to cause the at least one processor to operate so as to identify the total number of electronic files in the file storage and determine that the total number of electronic files exceeds a threshold amount.

15. The system of claim 10 , wherein the one or more processors are further configured to identify the total number of electronic files in the file storage and determine that the total number of electronic files exceeds a threshold amount.

16. The method of claim 1 , wherein the safeguard further includes flagging unanalyzed electronic files within the file storage as having a probability of including sensitive information.

17. The article of claim 6 , wherein the safeguard further includes flagging unanalyzed electronic files within the file storage as having a probability of including sensitive information.

18. The system of claim 10 , wherein the safeguard further includes flagging unanalyzed electronic files within the file storage as having a probability of including sensitive information.