Facilitating information extraction via semantic abstraction

• US 10,019,437 B2
• Filed: 02/23/2015
• Issued: 07/10/2018
• Est. Priority Date: 02/23/2015
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Ho, Ching-Tien, Kimelfeld, Benny, Li, Yunyao, Vaithyanathan, Shivakumar
• Primary Examiner(s): SPOONER, LAMONT M
• Application Number:
  US14/629,318
• Publication Number:
  US 20160246779A1
• Time in Patent Office: 1,233 Days
• Field of Search: 704 1, 704 9, 704 10
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

A method includes receiving one or more natural language dependency parse trees as input. A hardware processor is used for processing the dependency parse trees by creating a mapping from nodes of the one or more dependency parse trees into actions, roles and contextual predicates. The mapping is used for information extraction. The actions include the verbs along with attributes of the verbs. The roles include arguments for the verbs. The contextual predicates include modifiers for the verbs.

Claims

1. A method comprising:

receiving natural language dependency parse trees as input;

processing, using a hardware processor, the dependency parse trees using an action application program interface (API) that creates a mapping from nodes of the dependency parse trees into actions, roles and contextual predicates, eliminates auxiliary verbs, detects negation from adverbs, classifies verbs, and outputs a collection of annotation query language (AQL) views, wherein mappings other than a default mapping use a configuration file that provides for defining new roles through navigation in the dependency parse trees; and

using the mapping for information extraction,

wherein;

the actions are placed in a data format that comprises columns including said verbs along with columns for said verb attributes that comprise action identification, verb base, verb classification, verb tense, verb voice and verb sentiment, wherein verb sentiment is independent of an action;

the roles comprise arguments for said verbs; and

the contextual predicates include modifiers for said verbs.

2. The method of claim 1, wherein said processing further comprises:

identifying verbs and extracting verb attributes; and

associating the roles and context with said verbs.

3. The method of claim 2, wherein said processing further comprises:

identifying verb enclosures;

eliminating auxiliary verbs;

updating one or more AQL views for the input; and

outputting a collection of AQL views,

wherein an AQL view is a logical statement that defines a set of tuples, and all tuples in a view have a same schema.

4. The method of claim 1, wherein the roles further comprise lemmata and determiners for said verbs, and the modifiers comprise one or more of adverbs, temporal modifiers, and location modifiers.

5. The method of claim 4, wherein the AQL is used for defining the mapping of the dependency parse trees into the roles.

6. The method of claim 1, wherein for said processing, indices for verb categorization is used to determine the role names.

7. The method of claim 1, wherein said processing further comprises:

eliminating optional information from auxiliary verbs that is propagated into the action; and

exposing enclosure relationships among the actions.

8. The method of claim 1, wherein adverbs and explicit negation language are uniformly exposed as negative and semi-negative voice, and the contextual predicates are classified into types.

9. A computer program product for information extraction via semantic abstraction, the computer program product comprising a non-transitory computer readable storage medium having program code embodied therewith, the program code executable by a processor to:

receive natural language dependency parse trees as input;

process, using the processor, the dependency parse trees by using an action application program interface (API) that creates a mapping from nodes of the dependency parse trees into actions, roles and contextual predicates, eliminates auxiliary verbs, detects negation from adverbs, classifies verbs, and outputs a collection of annotation query language (AQL) views, wherein mappings other than a default mapping use a configuration file that provides for defining new roles through navigation in the dependency parse trees; and

using the mapping for extracting information,

wherein;

the actions are placed in a data format that comprises columns including said verbs along with columns for said verb attributes that comprise action identification, verb base, verb classification, verb tense, verb voice and verb sentiment, wherein verb sentiment is independent of an action;

the roles comprise arguments for said verbs; and

the contextual predicates include modifiers for said verbs.

10. The computer program product of claim 9, wherein said process further comprises:

identifying verbs and extracting verb attributes; and

associating the roles and context with said verbs.

11. The computer program product of claim 10, wherein said process further comprises:

identifying verb enclosures;

updating one or more AQL views for the input; and

outputting a collection of AQL views,

wherein an AQL view is a logical statement that defines a set of tuples, and all tuples in a view have a same schema.

12. The computer program product of claim 9, wherein the roles further comprise lemmata and determiners for said verbs, and the modifiers comprise one or more of adverbs, temporal modifiers, and location modifiers.

13. The computer program product of claim 12, wherein the AQL is used for defining the mapping of the dependency parse trees into the roles.

14. The computer program product of claim 9, wherein for said processing, indices for verb categorization is used to determine the role names.

15. The computer program product of claim 9, wherein said process further comprises:

eliminating optional information from auxiliary verbs that is propagated into the action; and

exposing enclosure relationships among the actions.

16. The computer program product of claim 9, wherein adverbs and explicit negation language are uniformly exposed as negative and semi-negative voice, and the contextual predicates are classified into types.

17. A system comprising:

a processor;

a storage device coupled to the processor, wherein the storage device stores one or more text files;

a parser that parses text into natural language dependency parse trees;

an action process that uses the processor for processing the dependency parse trees by using an action application program interface (API) that creates a mapping from nodes of the dependency parse trees into actions, roles and contextual predicates, eliminates auxiliary verbs, detects negation from adverbs, classifies verbs, and outputs a collection of annotation query language (AQL) views, wherein;

mappings other than a default mapping use a configuration file that provides for defining new roles through navigation in the dependency parse trees;

the mapping is used for information extraction;

the actions are placed in a data format that comprises columns including said verbs along with columns for said verb attributes that comprise action identification, verb base, verb classification, verb tense, verb voice and verb sentiment, wherein verb sentiment is independent of an action;

the roles comprise arguments for said verbs; and

the contextual predicates include modifiers for said verbs.

18. The system of claim 17, wherein said action process further comprises:

identifying verbs and extracting verb attributes;

associating the roles and context with said verbs;

identifying verb enclosures;

updating one or more AQL views for the input; and

outputting a collection of AQL views,

wherein an AQL view is a logical statement that defines a set of tuples, and all tuples in a view have a same schema.

19. The system of claim 17, wherein the roles further comprise lemmata and determiners for said verbs, the modifiers comprise one or more of adverbs, temporal modifiers, and location modifiers, for said processing, indices for verb categorization is used to determine the role names, and the AQL is used for defining the mapping of the dependency parse trees into the roles.

20. The system of claim 17, wherein said action process further comprises:

eliminating optional information from auxiliary verbs that is propagated into the action; and

exposing enclosure relationships among the actions,

wherein adverbs and explicit negation language are uniformly exposed as negative and semi-negative voice, and the contextual predicates are classified into types.