INBOUND CONTENT FILTERING VIA AUTOMATED INFERENCE DETECTION

• US 20090150365A1
• Filed: 12/05/2007
• Published: 06/11/2009
• Est. Priority Date: 12/05/2007
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Palo Alto Research Center, Inc. (Xerox Holdings Corp.)
• Inventors: Staddon, Jessica N., Chow, Richard, Golle, Philippe J.P.
• Granted Patent: US 7,860,885 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation

Abstract

A system is provided to detect email spam. During operation, the system receives an email, extracts a set of keywords from the email body, and constructs a first search query based a keyword extracted from the email body. The system further constructs a second search query based on the keyword in the first query and one additional word which pertains to a known spam word or to the subject of the email. Next, the system receives a first number of hits and a second number of hits in response to the first and second search queries, respectively. The system then determines whether the email is spam based on the first number and the second number. The system can also perform Website filtering using inference detection which is based on search results received in response to search queries formulated with keywords extracted from Websites.

Claims

1. A computer-executed method for filtering outbound content via inference detection, the method comprising:

identifying previously sent content with a common first destination identifier;

extracting a set of keywords from the identified previously sent content;

issuing a number of Web search queries based on the keywords extracted from the identified content;

receiving a number of hits in response to the queries;

extracting a set of expected-content keywords from the hits;

associating the expected-content keywords with the first destination identifier;

receiving a piece of outbound content;

determining the number of expected-content keywords present in the outbound content; and

producing a result which indicates the destination identifier for the outbound content and at least one expected-content keyword present in the outbound content, thereby allowing a user to determine whether the outbound content has the proper content and destination identifier.

2. The method of claim 1, wherein extracting the expected-content keywords from the hits comprises ranking the expected-content keywords.

3. The method of claim 1, wherein extracting the expected-content keywords from the hits comprises determining a term-frequency inverse-document-frequency (TF.IDF) weight for a respective word or phrase contained in a respective hit document.

4. The method of claim 1, wherein the method further comprises:

determining whether the outbound content'"'"'s destination identifier matches the first destination identifier.

5. The method of claim 4,

wherein producing the result comprises issuing a warning to the user when:

the outbound content'"'"'s destination identifier matches the first destination identifier and the number of expected-content keywords present in the outbound content is below a predetermined threshold;

or

when the outbound content'"'"'s destination identifier does not match the first destination identifier and the number of expected-content keywords present in the outbound content is above a predetermined threshold.

6. A computer-executed method for filtering outbound content via inference detection, the method comprising:

receiving a piece of outbound content;

extracting keywords from the outbound content;

issuing Web queries based on the keywords;

extracting keywords from documents corresponding to the hits returned by a search engine in response to the queries;

ranking the keywords extracted from the hits;

selecting one or more ranked keywords to represent the outbound content; and

producing a result which indicates a destination identifier for the outbound content and at least one keyword which is selected to represent the outbound content, thereby allowing a user to determine whether the outbound content has the proper content and destination identifier.

7. The method of claim 6, wherein extracting keywords from the outbound content comprises determining a TF.IDF weight for a respective word or phrase contained in the outbound content.

8. The method of claim 6, wherein extracting keywords from the hit documents comprises determining a TF.IDF weight for a respective word or phrase contained in a respective document.

9. The method of claim 6, wherein ranking the keywords extracted from the hit documents comprises determining the number of hit documents which contain a respective keyword.

10. The method of claim 6, wherein ranking the keywords extracted from the hit documents comprises determining the number of queries which result in at least one hit document that contains a respective keyword.

11. A computer system for filtering outbound content via inference detection, the computer system comprising:

a processor;

a memory coupled to the processor;

an content-selection mechanism configured to identify previously sent content with a common first destination identifier;

a keyword-extraction mechanism configured to extract a set of keywords from the identified previously sent content;

a query mechanism configured to;

issue a number of Web search queries based on the keywords extracted from the identified content;

receive a number of hits in response to the queries;

extract a set of expected-content keywords from the hits; and

associate the expected-content keywords with the first destination identifier;

a receiving mechanism configured to receive a piece of outbound content;

a decision mechanism configured to determine the number of expected-content keywords present in the outbound content; and

a display mechanism configured to produce a result which indicates the destination identifier for the outbound content and at least one expected-content keyword present in the outbound content, thereby allowing a user to determine whether the outbound content has the proper content and destination identifier.

12. The computer system of claim 11, wherein the query mechanism is further configured to rank the expected-content keywords.

13. The computer system of claim 11, wherein while extracting the expected-content keywords from the hits, the query mechanism is configured to determining a term-frequency inverse-document-frequency (TF.IDF) weight for a respective word or phrase contained in a respective hit document.

14. The computer system of claim 11, wherein the decision mechanism is further configured to:

determine whether the outbound content'"'"'s destination identifier matches the first destination identifier.

15. The computer system of claim 14,

wherein while producing the result, the display mechanism is configured to issue a warning to the user when:

the outbound content'"'"'s destination identifier matches the first destination identifier and the number of expected-content keywords present in the outbound content is

below a predetermined threshold;

or

when the outbound content'"'"'s destination identifier does not match the

first destination identifier and the number of expected-content keywords present in the outbound

content is above a predetermined threshold.

16. A computer system for filtering outbound content via inference detection, the computer system comprising:

a processor;

a memory coupled to the processor;

a receiving mechanism configured to receive a piece of outbound content;

a first keyword-extraction mechanism configured to extract keywords from the output content;

a query formulation mechanism configured to issue Web queries based on the keywords;

a second keyword-extraction mechanism configured to extract keywords from documents corresponding to the hits returned by a search engine in response to the queries;

a ranking mechanism configured to rank the keywords extracted from the hits;

a selection mechanism configured to select one or more ranked keywords to represent the outbound content; and

a display mechanism configured to produce a result which indicates a destination identifier for the outbound content and at least one keyword which is selected to represent the outbound content, thereby allowing a user to determine whether the outbound content has the proper content and destination identifier.

17. The computer system of claim 16, wherein while extracting keywords from the outbound content, the first keyword-extraction mechanism is configured to determine a TF.IDF weight for a respective word or phrase contained in the outbound content.

18. The computer system of claim 16, wherein while extracting keywords from the hit documents, the second keyword-extraction mechanism is configured to determine a TF.IDF weight for a respective word or phrase contained in a respective document.

19. The computer system of claim 16, wherein while ranking the keywords extracted from the hit documents, the ranking mechanism is configured to determine the number of hit documents which contain a respective keyword.

20. The computer system of claim 16, wherein while ranking the keywords extracted from the hit documents, the ranking mechanism is configured to determine the number of queries which result in at least one hit document that contains a respective keyword.