Summarizing text documents by resolving co-referentiality among actors or objects around which a story unfolds

• US 6,185,592 B1
• Filed: 11/18/1997
• Issued: 02/06/2001
• Est. Priority Date: 11/18/1997
• Status: Expired due to Term

• Current Assignee: Apple Inc.
• Original Assignee: Apple Computer Incorporated (Apple Inc.)
• Inventors: Kennedy, Christopher David, Boguraev, Branimir
• Primary Examiner(s): Hong, Stephen S.
• Assistant Examiner(s): Paula, Cesar B.
• Application Number:
  US08/974,079
• Time in Patent Office: 1,176 Days
• Field of Search: 707/530, 707/531, 704/9
• US Class Current: 715/256
• CPC Class Codes: G06F 40/258 Heading extraction; Automatic titling; Numbering

Abstract

A method and system for characterizing the content of a document is disclosed. The method and system comprise identifying a plurality of discourse referents in the document, dividing the document into topically relevant document segments, and resolving co-referentiality among the discourse referents within, and across, the document segments. The method and system also comprises calculating salience values for the discourse referents based upon the resolving step, and determining topic stamps for the document segments based upon the salience values of the associated discourse referents. Finally the method and system comprise providing summary-like abstractions, in the form of capsule overviews of each of the segments derived from its topic stamps. In so doing, a capsule overview is derived for the entire document, which will depict the core content of an average length article in a more accurate and representative manner than utilizing conventional techniques.

Claims

1. A method for characterizing the content of a document comprising the steps of:

a) identifying a plurality of discourse referents in the document;

b) dividing the document into topically relevant document segments;

c) resolving co-referentiality among the discourse referents within, and across, the document segments;

d) calculating salience values for the discourse referents based upon the resolving step;

e) determining topic stamps for the document segments based upon discourse salience values of the associated discourse referents; and

f) providing a capsule overview of the document, constructed from the topic stamps.

2. The method of claim 1 in which step (a) comprises the step of generating a plurality of discourse referents from a plurality of phrasal units.

3. The method of claim 1 in which the resolving step (c) comprises linking the discourse referents by co-referentiality with each other to assess a frequency with which they appear within a document and to establish prominence.

4. The method of claim 3 in which the salience calculating step (d) comprises evaluating a predetermined number of salience factors associated with each discourse referent.

5. The method of claim 4 in which the salience calculating step (d) assesses a plurality of anaphoric expressions on a basis of a shallow syntactic base alone.

6. The method of claim 5 wherein co-referentiality is determined primarily by resolving anaphoric expressions, and is represented by grouping discourse referents into equivalence classes.

7. The method of claim 6 in which the calculating step (d) comprises calculating a local salience value for each of the discourse referents and calculating a discourse salience value on the basis of co-reference classes of discourse referents.

8. The method of claim 7 in which the local salience value is the sum of the values of the salience factors that are satisfied by some member of the equivalence class to which a candidate antecedent belongs.

9. The method of claim 7 in which the discourse salience is defined to be a measure of the prominence of discourse referents across the document.

10. A system for characterizing the content of a document comprising:

means for identifying a plurality of discourse referents in the document;

means for dividing the document into topically relevant document segments;

means for resolving co-referentiality among the discourse referents within, and across, the document segments;

means for calculating salience values for the discourse referents based upon the resolving step;

means for determining topic stamps for the document segments based upon discourse salience values of the associated discourse referents; and

means for providing a capsule overview of the document, constructed from the topic stamps.

11. The system of claim 10 in which the identifying means comprises means for generating a plurality of discourse referents from a plurality of phrasal units.

12. The system of claim 10 in which the resolving means comprises means for linking the discourse referents by co-referentiality with each other to assess a frequency with which they appear within a document and to establish prominence.

13. The system of claim 12 in which the salience calculating means comprises means for evaluating a predetermined number of salience factors associated with each discourse referent.

14. The system of claim 13 in which the salience calculating means assesses a plurality of anaphoric expressions on a basis of a shallow syntactic base alone.

15. The system of claim 14 wherein co-referentiality is determined primarily by resolving anaphoric expressions, and is represented by grouping discourse referents into equivalence classes.

16. The system of claim 15 in which the calculating means comprises means for calculating a local salience value for each of the discourse referents and calculating a discourse salience value on the basis of co-reference classes of discourse-referents.

17. The system of claim 16 in which the local salience value is the sum of the values of the salience factors that are satisfied by some member of the equivalence class of which a candidate antecedent belongs.

18. The system of claim 16 in which the discourse salience is defined to be a measure of the prominence of discourse referents across the document.

19. A method for characterizing the content of a document comprising the steps of:

a) identifying a plurality of discourse referents in the document;

b) dividing the document into topically relevant document segments;

c) resolving co-referentiality among the discourse referents within, and across, the document segments, wherein the resolving step comprises linking the discourse referents by co-referentiality with each other to assess a frequency with which they appear within a document and to establish prominence;

d) calculating salience values for the discourse referents based upon the resolving step;

e) determining topic stamps for the document segments based upon discourse salience values of the associated discourse referents; and

f) providing a capsule overview of the document, constructed from the topic stamps.

20. A system for characterizing the content of a document comprising:

means for identifying a plurality of discourse referents in the document;

means for dividing the document into topically relevant document segments;

means for resolving co-referentiality among the discourse referents within, and across, the document segments wherein the resolving means comprises means for linking the discourse referents by co-referentiality with each other to assess a frequency with which they appear within a document and to establish prominence;

means for calculating salience values for the discourse referents based upon the resolving step;

means for determining topic stamps for the document segments based upon discourse salience values of the associated discourse referents; and

means for providing a capsule overview of the document, constructed from the topic stamps.