Method of utilizing implicit references to answer a query

• US 6,957,213 B1
• Filed: 05/17/2000
• Issued: 10/18/2005
• Est. Priority Date: 05/17/2000
• Status: Expired due to Term

• Current Assignee: Oracle OTC Subsidiary LLC (Oracle Corporation)
• Original Assignee: Inquira Incorporated (Oracle Corporation)
• Inventors: Yuret, Deniz
• Primary Examiner(s): Robinson, Greta
• Assistant Examiner(s): LEWIS, CHERYL RENEA
• Application Number:
  US09/572,770
• Time in Patent Office: 1,980 Days
• Field of Search: 707 1- 4, 704/1, 704/9, 704/10
• US Class Current: 707/711
• CPC Class Codes: G06F 16/90332 Natural language query formulation or dialogue systems
• CPC Class Codes: Y10S 707/99932 Access augmentation or optimizing
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation

Abstract

A method of utilizing implicit references to answer a query includes receiving segments of text, wherein individual segments have elements. Implicit references are inferred between elements of the segments. The inferring operation includes identifying implicit references to antecedent elements. A query is received. In response to the query, one or more segments are identified as relevant to the query based at least in part on the implicit references.

Claims

1. A method, comprising:

parsing text from at least one source of free-text into segments that each comprise elements;

inferring implicit references between elements of said segments prior to any query, wherein inferring includes identifying one or more implicit references to at least one antecedent element to form one or more identified implicit references;

receiving a query;

in response to said query, identifying one or more segments as relevant to said query based at least in part on said one or more implicit references; and

selecting said one or more segments for presentation.

2. The method of claim 1 wherein said parsing segments of text includes dividing said free-text source into said segments of text and corresponding generalizations.

3. The method of claim 1 in which said segments comprise sentences.

4. The method of claim 1 in which segments that are presented to said user are determined based on scoring.

5. The method of claim 1 in which only one segment is displayed.

6. The method of claim 1 in which only a single segment from a source is displayed.

7. The method of claim 1 wherein an implicit reference of said implicit references comprises a name variation that refers to an element.

8. The method of claim 1 wherein an implicit reference of said implicit references comprises an indirect reference to an element.

9. The method of claim 1 wherein an implicit reference of said implicit references comprises a pronoun.

10. The method of claim 1 wherein an implicit reference of said implicit references comprises a definite noun phrase.

11. The method of claim 1 wherein inferring implicit references further comprise deriving one or more generalized representations for said at least one antecedent element.

12. The method of claim 11 wherein deriving comprises:

generating said one or more generalized representations that refer to said at least one antecedent element; and

identifying each of said one or more generalized representations that refers to said at least one antecedent element as a potential reference.

13. The method of claim 12 wherein said one or more generalized representations include one or more of:

a number of name variations,

a number of single words,

a number of multi-word phrases,

a number of definite noun phrases, and

a number of indirect references.

14. The method of claim 12 further comprising:

associating said potential reference with one of said segments from which said at least one antecedent element was parsed; and

storing said one of said segments as a stored segment.

15. The method of claim 14 wherein storing comprises tagging said stored segment with said potential reference to form a tagged, stored segment, which is searchable by said potential reference.

16. The method of claim 12 wherein inferring implicit references further comprises:

generating alternative representations for an element other than said at least one antecedent element; and

matching at least one of said alternative representations with said potential reference to form one of said one or more identified implicit references.

17. The method of claim 16 further comprising:

associating said one of said one or more identified implicit references with said at least one antecedent element; and

supplementally storing another of said segments from which said element was parsed as a stored another segment.

18. The method of claim 17 wherein supplementally storing another of said segments comprises tagging said stored another segment with said at least one antecedent element to form a tagged, another stored segment, which is searchable by said at least one antecedent element.

19. The method of claim 18 wherein receiving said query further comprises:

parsing query text into query segments, each of which comprise query elements;

matching at least one query element with said at least one antecedent element that identifies said tagged, another stored segment; and

retrieving said tagged, another stored segment.

20. The method of claim 19 wherein retrieving said tagged, another stored segment further comprises searching an indexed database using said at least one antecedent element.

21. The method of claim 19 further comprising determining the relevancy of tagged, another stored segment to said query text.

22. A method, comprising:

extracting a first segment of text and a second segment of text, both of which are from an electronic document of free-text;

decomposing each of said first segment and said second segment into a first set of elements and a second set of elements, respectively;

inferring a number of implicit references between an antecedent element of said first set and a reference element of said second set prior to any query;

storing said second segment in a database such that said second segment is indexed by said antecedent element;

receiving a query comprising at least one query element; and

in response to said query, identifying at least one of said first segment or said second segment as relevant to said query based at least in part on one of said number of implicit references.

23. The method of claim 22 further comprising supplementally inferring a number of implicit query references based on potential query references derived from an antecedent query element of a preceding query.

24. The method of claim 23 wherein supplementally inferring said number of implicit query references further comprises deriving one or more generalized representations of said antecedent query element.

25. The method of claim 24 wherein deriving comprises:

generating said one or more generalized representations that refer to said antecedent query element; and

identifying each of said one or more generalized representations that refers to said antecedent query element as a potential query reference.

26. The method of claim 25 wherein identifying further comprises storing said antecedent query element and said one or more generalized representations in a first column and a corresponding second column, respectively, wherein said first column and said second column constitute a short term buffer.

27. The method of claim 25 wherein supplementally inferring said number of implicit query references further comprises matching said query element with said potential query reference to form a matched query element, which is one of said number of implicit query references.

28. The method of claim 27 wherein identifying one or more segments as relevant further comprises retrieving said second segment using said one of said number of implicit query references to search a database.