SEMANTIC COMPATIBILITY CHECKING FOR AUTOMATIC CORRECTION AND DISCOVERY OF NAMED ENTITIES

• US 20090204596A1
• Filed: 02/08/2008
• Published: 08/13/2009
• Est. Priority Date: 02/08/2008
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Hagege, Caroline, Brun, Caroline
• Granted Patent: US 8,000,956 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 40/295 Named entity recognition

Abstract

A computer implemented system and method for processing text are disclosed. Partially processed text, in which named entities have been extracted by a standard named entity system, is processed to identify attributive relations between a named entity or proper noun and a corresponding attribute. A concept for the attribute is identified and, in the case of a named entity, compared with the named entity'"'"'s context, enabling a confirmation or conflict between the two to be determined. In the case of a proper name, the attribute'"'"'s context can be associated with the proper name, allowing the proper name to be recognized as a new named entity.

Claims

1. A method of processing text comprising:

receiving processed text which has original information associated with it identifying text elements corresponding to named entities and their assigned contexts, the text comprising a text string;

for the text string, applying at least one grammar rule for identifying an attributive relationship between an attribute and a nominal entity in the text string, the nominal entity comprising at least one of;

an identified named entity having an assigned context, and

an identified proper name which is not identified in the original information as being a named entity;

identifying at least one concept for the attribute;

where the nominal entity is a named entity, comparing the assigned named entity context with the at least one attribute context; and

providing for new information to be associated with the text, which is based on at least one of;

the comparison, and

the identified context of an attribute which is identified as being in an attributive relationship with an identified proper name.

2. The method of claim 1, wherein when the entity is a named entity and the at least one attribute context being compared is a single attribute context, the comparison includes determining whether the entity context and attribute context are compatible.

3. The method of claim 2, wherein when the named entity context and the at least one attribute context are determined not to be compatible, the comparison identifies a conflict.

4. The method of claim 3, wherein the new information to be associated with the text comprises a correction to the identified context for the named entity or evidences the existence of the conflict.

5. The method of claim 3, wherein the providing includes identifying the conflict to a user for manual resolution of the conflict.

6. The method of claim 3, further comprising, wherein when the attribute context is a person type of attribute, and the identified context of the entity with which there is a conflict is one of an organization type of context and a location type of context, identifying that entity as having a metonymic use in the text string.

7. The method of claim 1, wherein the at least one attribute context comprises a plurality of attribute contexts and wherein the named entity context and attribute contexts of a named entity and attribute which are identified as being in an attributive relation are determined not to be compatible only when none of the attribute contexts is compatible with the named entity context.

8. The method of claim 1, wherein when the nominal entity is a proper name, the additional information to be associated with the text identifies the proper name as being a discovered named entity.

9. The method of claim 1, wherein the identifying of at least one context for the attribute includes retrieving the at least one context from a lexical ontology in which attributes are indexed based on context.

10. The method of claim 1, wherein the text comprises a plurality of text strings and the method further includes propagating the new information from one of the text strings to another of the text strings.

11. The method of claim 1, wherein the least one grammar rule comprises a set of grammar rules each of which is designed for identifying a loose ISA relationship.

12. The method of claim 1, wherein each of the at least one attribute contexts is a different context from a set of contexts which are all associated with nouns in the lexical ontology.

13. The method of claim 1, further comprising:

prior to receiving of the processed text, receiving text and processing the text with a named entity recognition system to identify the text elements corresponding to named entities and assigning respective contexts from a finite set of contexts.

14. The method of claim 1, further comprising:

annotating the text with additional information based on the comparison.

15. The method of claim 1, wherein the text is received from a named entity recognition system.

16. The method of claim 15, wherein the text comprises a plurality of text strings and the method further includes determining an accuracy of the named entity recognition system based on a set of the comparisons.

17. An apparatus comprising memory which stores instructions for performing the method of claim 1 and a processor, in communication with the memory for executing the instructions.

18. A computer program product comprising tangible media which encodes instructions which, when executed by a computer, perform the method of claim 1.

19. An apparatus for processing text comprising:

an attribute identification component which, for a text string of received text which has information associated with it identifying text elements corresponding to named entities and their assigned contexts, applies at least one grammar rule for identifying an attributive relationship between an attribute and a nominal entity in the text string, the nominal entity comprising at least one of an identified named entity and an identified proper name which has not been identified as a named entity;

a compatibility checker which accesses an associated lexical ontology to identify at least one concept for the attribute that is identified as being in an attributive relationship with the entity and, in the case of a named entity, compares the at least one attribute context with the assigned context of the entity; and

an annotation component which annotates the text, based on at least one of;

the comparison, and

the identified context of an attribute which is identified as being in an attributive relationship with an identified proper name.

20. The apparatus of claim 19, further comprising a named entity recognition component which identifies text elements corresponding to named entities and assigns a context to each of the identified named entities form a set of contexts.

21. A method of processing text comprising:

processing text to form processed text which identifies named entities in the text and their corresponding contexts;

for at least one text string in the processed text, applying at least one grammar rule for identifying an attributive relationship between an attribute and a nominal entity in the text string, the nominal entity being at least one of;

one of the identified named entities, and

an identified proper name which is not identified in the processed text as being a named entity;

identifying at least one concept for the attribute;

where the nominal entity is a named entity, comparing the named entity'"'"'s context with the at least one attribute context; and

where the nominal entity is either a proper name or there is a conflict between the compared named entity'"'"'s context with the at least one attribute context, associating information with the processed text based thereon.