# US 2013/0297626 A1 · Systems and methods for extraction of policy information

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2013/0297626 A1
- Document ID: US20130297626A1
- Status: Published application
- Kind: A1
- Application number: 13/826,776
- Filed: 2013-03-14
- Published: 2013-11-07
- Earliest claimed: 2012-03-23
- Assignees: Avast Software B.V.
- Inventors: Shaul Levi; Valery Kholodkov; Yuval Ben-Itzhak
- Web page: https://patentgrep.com/patent/US20130297626A1

## CPC

- G06F 16/353: into predefined classes
- G06F 16/21: Design, administration or maintenance of databases
- G06F 16/345: {Summarisation for human users}
- G06F 18/24147: {Distances to closest patterns, e.g. nearest neighbour classification}
- G06F 18/24155: {Bayesian classification}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

In a system for extracting policy information from text, a processor analyzes if the text is relevant to a top-level category, and then determines if at least a portion of the text is relevant to categories and subcategories within a taxonomy of categories and subcategories related to the top-level category. If at least a portion of the text is determined to be relevant to the category/subcategory, a classifier extracts policy information associated with the category/subcategory. Using text that includes a known policy the classifiers can be trained to correctly recognize categories/subcategories, and the values associated therewith.

## Claims

### Claim 1

A method of extracting policy information from text, the method comprising:
analyzing, by a first processor, if the text is relevant to a top-level category;
if the text is determined to be relevant to the top-level category, determining by the first processor if at least a portion of the text is relevant to a category within a taxonomy of categories related to the top-level category; and
if at least a portion of the text is determined to be relevant to the category, extracting from the relevant portion of the text, using a first classifier, policy information associated with the category.

### Claim 2 (depends on 1)

The method of claim 1, further comprising extracting from the relevant portion of the text, using a second classifier, policy information associated with a subcategory within the taxonomy, the subcategory being related to the category.

### Claim 3 (depends on 1)

The method of claim 1, wherein the text comprises one of a sentence and a document.

### Claim 4 (depends on 1)

The method of claim 1, wherein at least one of the analyzing, determining, and extracting steps comprises machine learning by the first processor.

### Claim 5 (depends on 1)

The method of claim 1, wherein wherein at least one of the analyzing, determining, and extracting steps comprises comprises natural language processing.

### Claim 6 (depends on 1)

The method of claim 1, wherein the first classifier employs statistical classification.

### Claim 7 (depends on 6)

The method of claim 6 wherein the statistical classification comprises at least one of multinomial logistic regression, word profile similarity comparison, naive Bayes classification, k-nearest neighbor classification, and maximal likelihood based classification.

### Claim 8 (depends on 1)

The method of claim 1, wherein the top-level category is selected from the group consisting of a privacy policy, a lease agreement, and a license.

### Claim 9 (depends on 1)

The method of claim 1, wherein the category is selected from the group consisting of collecting personally-identifiable information, collecting non-personally-identifiable information, sharing collected information, and allowing opt-out.

### Claim 10 (depends on 9)

The method of claim 9, wherein the subcategory is related to collecting personally-identifiable information, and the subcategory is selected from the group consisting of collection of name, collection of address, collection of email address, and collection of IP address.

### Claim 11 (depends on 1)

The method of claim 1, further comprising:
storing the extracted policy information in a database; and
transmitting the stored policy information.

### Claim 12 (depends on 11)

The method of claim 11, wherein the database comprises at least one of a local database and a distributed database.

### Claim 13

A method of responding to policy information extracted from text, the method comprising:
receiving extracted policy information, the extraction being performed by:
(i) analyzing, by a first processor, if the text is relevant to a top-level category,
(ii) if the text is determined to be relevant to the top-level category, determining by the first processor if at least a portion of the text is relevant to a category within a taxonomy of categories related to the top-level category, and
(iii) if at least a portion of the text is determined to be relevant to the category, identifying from the relevant portion of the text, using a classifier, policy information associated with the category;
displaying, by a second processor, the received extracted policy information to a user;
receiving an instruction from the user in response to the displayed policy information; and
taking an action by the second processor in response to the received instruction.

### Claim 14 (depends on 13)

The method of claim 13, wherein the second processor is the first processor.

### Claim 15 (depends on 13)

The method of claim 13, wherein the action taken comprises limiting access to at least a part of a source associated with the text.

### Claim 16 (depends on 15)

The method of claim 15, wherein the source is selected from the group consisting of a website, an application, and a device.

### Claim 17 (depends on 13)

The method of claim 13, further comprising adjusting the classifier if at least one of the portion of the text determined to be relevant to the category and the extracted policy information is not related to the category.

### Claim 18

(canceled)

### Claim 19

(canceled)

### Claim 20

A system for extracting policy information from text, the system comprising:
a storage medium; and
a processor configured as:
(i) an analyzer adapted for: (a) analyzing if the text is relevant to a top-level category, and (b) determining, if the text is determined to be relevant to the top-level category, if at least a portion of the text is relevant to a category within a taxonomy of categories related to the top-level category, and
(ii) a first classifier adapted for extracting from the relevant portion of the text policy information associated with the category, if at least a portion of the text is determined to be relevant to the category.

### Claim 21 (depends on 20)

The system of claim 20, wherein the processor is further configured as a second classifier adapted for extracting from the relevant portion of the text policy information associated with a subcategory within the taxonomy, the subcategory being related to the category.

### Claim 22 (depends on 20)

The system of claim 20, wherein the text comprises one of a sentence and a document.

### Claim 23 (depends on 20)

The system of claim 20, wherein at least one of the analyzer and the first classifier is further adapted for machine learning.

### Claim 24 (depends on 20)

The system of claim 20, wherein at least one of the analyzer and the first classifier is further adapted for natural language processing.

### Claim 25 (depends on 20)

The system of claim 20, wherein the first classifier employs statistical classification.

### Claim 26 (depends on 25)

The system of claim 25, wherein the statistical classification comprises at least one of multinomial logistic regression, word profile similarity comparison, naive Bayes classification, k-nearest neighbor classification, and maximal likelihood based classification.

### Claim 27 (depends on 20)

The system of claim 20, wherein the top-level category is selected from the group consisting of a privacy policy, a lease agreement, and a license.

### Claim 28 (depends on 20)

The system of claim 20, wherein the category is selected from the group consisting of collecting personally-identifiable information, collecting non-personally-identifiable information, sharing collected information, and allowing opt-out.

### Claim 29 (depends on 28)

The system of claim 28, wherein the subcategory is related to collecting personally-identifiable information, and the subcategory is selected from the group consisting of collection of name, collection of address, collection of email address, and collection of IP address.

### Claim 30 (depends on 20)

The system of claim 20, wherein the processor is further configured for:
(iii) storing the extracted policy information in a database; and
(iv) transmitting the stored policy information.

### Claim 31 (depends on 30)

The system of claim 30, wherein the database comprises at least one of a local database and a distributed database.

### Claim 32 (depends on 20)

The system of claim 20, wherein the processor is further configured for:
displaying the extracted policy information to a user;
receiving an instruction from the user in response to the displayed policy information; and
taking an action in response to the received instruction.

### Claim 33 (depends on 32)

The system of claim 32, wherein the action taken comprises limiting access to at least a part of a source associated with the text.

### Claim 34 (depends on 33)

The system of claim 33, wherein the source is selected from the group consisting of a website, an application, and a device.

### Claim 35 (depends on 20)

The system of claim 20, wherein the processor is further configured for adjusting the first classifier if at least one of the portion of the text determined to be relevant to the category and the extracted policy information is not related to the category.

### Claim 36

(canceled)

### Claim 37

(canceled)
