Machine learning to determine analogy outcomes

• US 9,892,112 B1
• Filed: 10/06/2016
• Issued: 02/13/2018
• Est. Priority Date: 10/06/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Allen, Corville O., Freed, Andrew R.
• Primary Examiner(s): Pullias, Jesse
• Application Number:
  US15/287,113
• Time in Patent Office: 495 Days
• Field of Search: 704 1, 704 9, 704 10, 704257, 704270-275
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/295 Named entity recognition
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

Embodiments relate to a system, program product, and method for use with an intelligent computer platform to decipher analogical phrases. The aspect of deciphering the phrase includes parsing a phrase into subcomponents, identifying a category for each parsed subcomponent and a syntactic structure of the phrase, and generating a list of definitions for each parsed subcomponent. Definitions in the list are ranked according to relevance and an outcome base is identified based on ranked relevancy, with the outcome being a definition with the highest relevance in the list. A corpus is searched for evidence of a pattern associated with the list. Each definition in the list is scored according to a weighted calculation based on congruence of corpus evidence with the pattern. An outcome is generated from the pattern that receives the highest score based on congruence with the corpus evidence.

Claims

1. A computer system comprising:

a processing unit operatively coupled to memory;

an artificial intelligence platform, in communication with the processing unit;

a knowledge engine in communication with the processing unit to decipher a phrase upon activation by the artificial intelligence platform, including;

parse a phrase into subcomponents;

identify a category for each parsed subcomponent and a syntactic structure of the phrase;

generate a list of definitions for each parsed subcomponent, the list corresponding to the identified category;

rank the definitions in the list according to relevance;

identify an outcome base on ranked relevancy, the outcome being a definition with the highest relevance in the list;

search a corpus for evidence of a pattern associated with the list;

score each definition in the list according to a weighted calculation based on congruence of corpus evidence with the pattern; and

generate an outcome from the definition list, wherein the outcome is a definition with a strongest congruence to the pattern.

2. The computer system of claim 1, wherein the category is selected from the group consisting of:

noun, entity, subject, hypernym, hyponym, verb, comparator, idiom, and adjective.

3. The computer system of claim 1, wherein the parse of the phrase sub-components is performed in accordance to Slot Grammar Logic.

4. The computer system of claim 1, wherein each definition is an explanation of each parsed subcomponent.

5. The computer system of claim 1, wherein the definition rank is in accordance to relevance, and includes a filter of the definitions based on at least one similar adjective.

6. The computer system of claim 1, wherein the definition rank is in accordance to relevance, and includes a filter of the definitions based on at least one similar adverbs.

7. A computer program product to decipher a phrase, the computer program product comprising a computer readable storage device having program code embodied therewith, the program code executable by a processing unit to:

parse a phrase into subcomponents using a knowledge engine;

identify a category for each parsed subcomponent and a syntactic structure of the phrase;

generate a list of definitions for each parsed subcomponent, the list corresponding to the identified category;

rank the definitions in the list according to relevance;

identify an outcome base on ranked relevancy, the outcome being a definition with the highest relevance in the list;

search a corpus for evidence of a pattern associated with the list;

score each definition in the list according to a weighted calculation based on congruence of corpus evidence with the pattern; and

generate an outcome, wherein the outcome is a definition with a strongest congruence to the pattern.

8. The computer program product of claim 7, wherein the category is selected from the group consisting of:

noun, entity, subject, hypernym, hyponym, verb, comparator, idiom, and adjective.

9. The computer program product of claim 7, wherein the program code to parse a phrase into subcomponents is performed in accordance with Slot Grammar Logic.

10. The computer program product of claim 7, wherein each definition is an explanation of each parsed subcomponent.

11. The computer program product of claim 7, wherein the program code to rank the definitions in the list according to relevance includes the processing unit to filter the definitions based on similar adjectives.

12. The computer program product of claim 7, wherein the program code to rank the definitions in the list according to relevance includes the processing unit to filter the definitions based on similar adverbs.

13. A method employing a knowledge engine for processing natural language input, comprising:

parsing a phrase into subcomponents using the knowledge engine;

identifying a category for each parsed subcomponent and a syntactic structure of the phrase;

generating a list of definitions for each parsed subcomponent, the list corresponding to the identified category;

ranking the definitions in the list according to relevance;

identifying an outcome base on ranked relevancy, the outcome being a definition with the highest relevance in the list;

searching a corpus for evidence of a pattern associated with the list;

scoring each definition in the list according to a weighted calculation based on congruence of corpus evidence with the pattern; and

generating an outcome, wherein the outcome is a definition with a strongest congruence to the pattern.

14. The method of claim 13, wherein the category is selected from the group consisting of:

noun, entity, subject, hypernym, hyponym, verb, comparator, idiom, and adjective.

15. The method of claim 13, wherein the parsing is performed in accordance to Slot Grammar Logic.

16. The method of claim 13, wherein each definition is an explanation of each parsed subcomponent.

17. The method of claim 13, wherein the ranking according relevance includes filtering the definitions based on similar adjectives.

18. The method of claim 13, wherein the ranking according to relevance includes filtering the definitions based on similar adverbs.