Unsupervised extraction of facts

• US 8,825,471 B2
• Filed: 03/31/2006
• Issued: 09/02/2014
• Est. Priority Date: 05/31/2005
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Betz, Jonathan T., Zhao, Shubin
• Primary Examiner(s): SPOONER, LAMONT M
• Application Number:
  US11/394,414
• Publication Number:
  US 20070150800A1
• Time in Patent Office: 3,077 Days
• Field of Search: 704/1, 704/9, 704/10, 707 2- 6, 707706-708, 715/255
• US Class Current: 704/9
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06F 16/84 Mapping; Conversion
• CPC Class Codes: G06F 16/93 Document management systems
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: Y10S 707/962 Entity-attribute-value

Abstract

A system and method for extracting facts from documents. A fact is extracted from a first document. The attribute and value of the fact extracted from the first document are used as a seed attribute-value pair. A second document containing the seed attribute-value pair is analyzed to determine a contextual pattern used in the second document. The contextual pattern is used to extract other attribute-value pairs from the second document. The extracted attributes and values are stored as facts.

Claims

1. A computer-implemented method for extracting facts, the method comprising:

at a computer system including one or more processors and memory storing one or more programs, the one or more processors executing the one or more programs to perform the operations of;

identifying a predefined pattern in a first document;

extracting a first fact having an attribute and a value from the first document based on the predefined pattern;

retrieving a second document that contains the attribute and the value of the first fact;

identifying in the second document a contextual pattern associated with the attribute and value of the first fact;

extracting a second fact from the second document using the contextual pattern; and

storing the first fact and the second fact in a fact repository of the computer system.

2. The method of claim 1, further comprising:

extracting a third fact from a third document using the contextual pattern.

3. The method of claim 2, wherein the second document is hosted on a first domain, and wherein the third document is hosted on the first domain.

4. The method of claim 1, further comprising:

associating the first fact with a first object.

5. The method of claim 4, wherein the first object is associated with an object name, and wherein retrieving the second document comprises searching a repository of documents for a document containing the object name.

6. The method of claim 4, further comprising associating the second fact with the first object.

7. The method of claim 1, wherein extracting a first fact from a first document comprises:

extracting a first plurality of facts from the first document, each fact having an attribute and a value.

8. The method of claim 7, wherein identifying in the second document a contextual pattern associated with the attribute and value of the first fact comprises:

identifying in the second document a contextual pattern associated with the attributes and the values of a number of the first plurality of facts.

9. The method of claim 8, wherein said extracting said second fact is responsive to the number of the first plurality of facts having attributes and values associated with the contextual pattern.

10. The method of claim 7, wherein said first plurality of facts includes a third fact, the method further comprising:

determining if the third fact is organized in the second document according to the contextual pattern.

11. The method of claim 1, wherein said first document is different from said second document.

12. The method of claim 1, wherein retrieving the second document comprises querying a search engine for a document containing the attribute and the value of the first fact.

13. A system for extracting facts

one or more processors;

and

memory storing one or more programs to be executed by the one or more processors;

the one or more programs comprising instructions for;

identifying a predefined pattern in a first document, extract a first fact having an attribute and value from the first document based on the predefined pattern, and store the first fact in a fact repository; and

receiving said first fact, identify a contextual pattern of said first fact in a second document, and extract a second fact from the second document using the contextual pattern, and store the second fact in the fact repository.

14. The system of claim 13, further comprising:

a document repository, wherein the first document and the second document are stored in said document repository.

15. The system of claim 13, further comprising:

a fact repository configured to store the first fact and the second fact.

16. A non-transitory computer readable storage medium storing one or more programs configured for execution by a computer, the one or more programs comprising instructions for:

program code for identifying a predefined pattern in a first document;

extracting a first fact having an attribute and a value from the first document based on the predefined pattern;

retrieving a second document that contains the attribute and the value of the first fact;

identifying in the second document a contextual pattern associated with the attribute and value of the first fact;

extracting a second fact from the second document using the contextual pattern; and

storing the first fact and the second fact in a fact repository.

17. The non-transitory computer readable storage medium of claim 16, the computer-readable medium further comprising:

program code for extracting a third fact from a third document using the contextual pattern.

18. The non-transitory computer readable storage medium of claim 16, the computer-readable medium further comprising:

program code for associating the first fact with a first object.

19. The non-transitory computer readable storage medium of claim 18, wherein the first object is associated with an object name, and wherein the program code for retrieving the second document comprises program code for searching a repository of documents for a document containing the object name.

20. The non-transitory computer readable storage medium of claim 18, further comprising program code for associating the second fact with the first object.

21. The non-transitory computer readable storage medium of claim 16, wherein the computer code for retrieving the second document comprises computer code for querying a search engine for a document containing the attribute and the value of the first fact.

22. A computer-implemented method for extracting facts, the method comprising:

at a computer system including one or more processors and memory storing one or more programs, the one or more processors executing the one or more programs to perform the operations of;

identifying a predefined pattern in a first document;

extracting a first fact having an attribute and a value from the first document based on the predefined pattern;

retrieving a second document;

if the second document corroborates the first fact;

retrieving a third document that contains the attribute and the value of the first fact;

identifying in the third document a contextual pattern associated with the attribute and value of the first fact;

extracting a second fact from the third document using the contextual pattern; and

storing the first fact and the second fact in a fact repository of the computer system.

23. The method of claim 22, further comprising:

retrieving a fourth document;

if the fourth document corroborates the second fact, storing the second fact in a fact repository.