Language independent probabilistic content matching

• US 9,087,039 B2
• Filed: 02/07/2012
• Issued: 07/21/2015
• Est. Priority Date: 02/07/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Gandhi, Mauktik, Lamanna, Charles, Sankaranarayanan, Vidyaraman, Pontes Filho, Raimundo
• Primary Examiner(s): JACKSON, JAKIEDA R
• Application Number:
  US13/367,469
• Publication Number:
  US 20130204609A1
• Time in Patent Office: 1,260 Days
• Field of Search: 704/244, 704/9
• US Class Current: 1/1
• CPC Class Codes: G06F 21/6263 during internet communication, e.g. revealing personal data from cookies
• CPC Class Codes: G06F 40/166 Editing, e.g. inserting or deleting
• CPC Class Codes: G06F 40/20 Natural language analysis semantic analysis of natural language G06F40/30
• CPC Class Codes: G06F 40/279 Recognition of textual entities
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: H04L 63/10 for controlling access to devices or network resources

Abstract

Content is received and compared against rules for identifying a type of content. Each rule has both segmented and unsegmented patterns. The content is matched against the patterns and assigned a confidence score that is higher if the content matches a segmented pattern and lower if the content matches an unsegmented pattern.

Claims

1. A computer-implemented method comprising:

receiving electronic message a content source system, the electronic message including textual content;

accessing, from a rules store by a content determination component in a content determination system, a set of rules that define patterns that are used to identify content as sensitive content, each rule having segmented and un-segmented patterns that can be matched to textual content written in a segmented language and textual content written in an un-segmented language, respectively;

matching, by the content determination component, at least one textual content portion in the electronic message against the patterns in each rule, regardless of whether the at least one textual content portion in the electronic message is written in a segmented language or in an un-segmented language, to determine whether the at least one textual content portion in the electronic message is sensitive content;

generating, by the content determination component, a confidence score corresponding to the determination as to whether the at least one textual content portion in the electronic message is sensitive content, based on whether the at least one textual content portion in the electronic message matched a segmented pattern or an un-segmented pattern; and

applying, by a content processing system, a data dissemination rule to the at least one textual content portion in the electronic message based on the determination as to whether the at least one textual content portion in the electronic message is sensitive content and the corresponding confidence score, wherein applying the data dissemination, rule comprises at least one of;

blocking the at least one textual content portion in the electronic message from being sent to a potential recipient;

displaying a message indicating that the at least one textual content portion contains sensitive material and that the at least one textual content portion will be blocked from being sent to a potential recipient;

or

displaying a message indicating that the at least one textual content portion contains sensitive material and instructing the user how to proceed based on the data dissemination rule.

2. The computer-implemented method of claim 1 wherein generating a confidence score comprises:

generating, by the content determination component, a higher confidence score if the at least one textual content portion in the electronic message matched a segmented pattern and generating a lower confidence score if the at least one textual content portion in the electronic message matched an un-segmented pattern.

3. The computer-implemented method of claim 1 wherein applying a data dissemination rule to the at least one textual content portion, comprises:

if the at least one textual content portion in the electronic message is the first type of content with a sufficient corresponding confidence score, processing, by the content processing system, the at least one textual content portion in the electronic message according to a first set of processing rules; and

otherwise, processing, by the content processing system, the at least one textual content portion in the electronic message according to a second set of processing rules.

4. The computer-implemented method of claim 1 wherein matching comprises:

matching, by the content determination component, the at least one textual content portion in the electronic message against both the un-segmented patterns and the segmented patterns in each rule.

5. The computer-implemented method of claim 4 wherein each pattern includes a main match pattern and a corroborating match portion and wherein matching comprises:

matching, by the content determination component, the at least one textual content portion in the electronic message first against the main match portion of a given pattern;

if the at least one textual content portion in the electronic message matches the main matching portion of the given pattern, then matching, by the content determination component, the at least one textual content portion in the electronic message against the corroborating match portion; and

if the at least one textual content portion in the electronic message matches the corroborating match portion of the given pattern, then assigning a first confidence score to the match.

6. The computer-implemented method of claim 5 wherein the corroborating match portion of the given pattern has a plurality of corroborating patterns, and wherein matching the at least one textual content portion in the electronic message against the corroborating match portion comprises:

matching, by the content determination component, the at least one textual content portion in the electronic message against each of the plurality of corroborating match patterns; and

if the at least one textual content portion in the electronic message matches more than one corroborating match pattern, increasing the confidence score.

7. The computer-implemented method of claim 1, wherein the content source system and the content processing system comprise parts of a same system.

8. The computer-implemented method of claim 1, wherein the electronic message comprises an email message and the at least one textual content portion in the electronic message comprises at least one of:

an email body or an email attachment.

9. The computer-implemented method of claim 1 wherein matching comprises:

matching, by the content determination component, the at least one textual content portion in the electronic message against patterns in a rule that identify at least one of personal information, credit card information, or social security number information.

10. A content processing system, comprising:

a rules data store that includes a plurality of rules, each rule corresponding to a type of information and having a set of segmented patterns and a set of un-segmented patterns;

a content determination component receiving an electronic message having content and matching the content against the segmented patterns and the un-segmented patterns in each rule to determine whether the content includes the type of information corresponding to each rule, the content determination component assigning a confidence level to the determination of whether the content includes the type of information corresponding to a given rule based on whether the content matched a segmented pattern or an un-segmented pattern in the given rule;

a content processing system processing the content based on the determination of whether the content includes the type of information corresponding to the given rule, and applying a data dissemination rule comprises at least one of blocking the at least one textual content portion in the electronic message from being sent to a potential recipient, displaying a message indicating that the at least one textual content portion contains sensitive material and that the at least one textual content portion will be blocked from being sent to a potential recipient, or displaying a message indicating that the at least one textual content portion contains sensitive material and instructing the user how to proceed in accordance with the data dissemination rule; and

a computer processor being a functional component of the system and being activated by the content determination component to facilitate matching and assigning a confidence level.

11. The content processing system of claim 10 wherein the content determination component matches the content against the segmented patterns and the un-segmented patterns regardless of whether the content is written in a segmented language, an un-segmented language or both a segmented and an un-segmented language.

12. The content processing system of claim 11 wherein the content determination component assigns a higher confidence level to the determination if the content matched a segmented pattern than if the content matched an un-segmented pattern.

13. The content processing system of claim 12 wherein each pattern in a rule has a main matching portion and a corroborating matching portion and wherein the content determination component matches the content first against the main matching portion and then, if the content matched the main matching portion, against the corroborating matching portion.

14. The content processing system of claim 13 wherein the corroborating matching portion includes a plurality of matching patterns.

15. The content processing system of claim 14 wherein the content determination component assigns an increased confidence level if the content matches more than one of the plurality of matching patterns in the corroborating matching portion.

16. A non-volatile memory device having computer readable instructions which, when executed, perform a method, comprising:

receiving an electronic message from a content source system, the electronic message including textual content;

accessing, from a rules store by a content determination component in a content determination system, a set of rules that define patterns that are used to identify content as sensitive content, each rule having segmented and un-segmented patterns that can be matched to textual content written in a segmented language and textual content written in an un-segmented language, respectively;

matching, by the content determination component, at least one textual content portion in the electronic message against the patterns in each rule, regardless of whether the at least one textual content portion in the electronic message is written in a segmented language or in an un-segmented language, to determine whether the at least one textual content portion in the electronic message is sensitive content;

generating, by the content determination component, a confidence score corresponding to the determination as to whether the at least one textual content portion in the electronic message is sensitive content, based on whether the at least one textual content portion in the electronic message matched a segmented pattern or an un-segmented pattern by generating a higher confidence score if the at least one textual content portion in the electronic message matched a segmented pattern and generating a lower confidence score if the at least one textual content portion in the electronic message matched an un-segmented pattern; and

applying, by a content processing system, a data dissemination rule to the at least one textual content portion in the electronic message based on the determination as to whether the at least one textual content portion in the electronic message is sensitive content and the corresponding confidence score, wherein applying the data dissemination rule comprises at least one of;

blocking the at least one textual content portion in the electronic message from being sent to a potential recipient;

displaying a message indicating that the at least one textual content portion contains sensitive material and that the at least one textual content portion will be blocked from being sent to a potential recipient;

or

displaying a message indicating that the at least one textual content portion contains sensitive material and instructing the user how to comply with the data dissemination rule.

17. The non-volatile memory device of claim 16, wherein applying a data dissemination rule to the at least one textual content portion, comprises:

if the at least one textual content portion in the electronic message is the first type of content with a sufficient corresponding confidence score, processing, by the content processing system, the at least one textual content portion in the electronic message according to a first set of processing rules; and

otherwise, processing, by the content processing system, the at least one textual content portion in the electronic message according to a second set of processing rules.

18. The non-volatile memory device of claim 16, wherein matching comprises:

matching, by the content determination component, the at least one textual content portion in the electronic message against both the un-segmented patterns and the segmented patterns in each rule.

19. The non-volatile memory device of claim 16, wherein the electronic message comprises an email message and the at least one textual content portion in the electronic message comprises at least one of:

an email body or an email attachment.