Method for improving the accuracy of a statement

• US 10,657,204 B2
• Filed: 07/17/2018
• Issued: 05/19/2020
• Est. Priority Date: 07/17/2018
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Brake, Kyle M., Boxwell, Stephen A., Frost, Keith G., Vernier, Stanley J.
• Primary Examiner(s): Neway, Samuel G
• Application Number:
  US16/037,711
• Publication Number:
  US 20200026754A1
• Time in Patent Office: 672 Days
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique
• CPC Class Codes: G06F 40/279 Recognition of textual entities
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition

Abstract

A cognitive assistant system, computer program product, and computer-implemented method configured to improve the accuracy of a statement. In one example, the system receives from a user an original statement that includes statement entities and generates an original confidence estimate (CE) for a correctness of the original statement. The system generates a replacement statement by selecting a suspect statement entity, selecting a replacement entity for the suspect statement entity, and replacing the suspect statement entity with the replacement entity in the original statement. The system generates a replacement CE for a correctness of the replacement statement. If the replacement CE exceeds the original CE by a tuning factor, the system sends the replacement statement to the user. If not, the system repeats generating the replacement statement and generating the replacement CE.

Claims

1. A computer-implemented method performed by a cognitive assistant (CA) system for improving an accuracy of a statement, comprising:

receiving from a user an original statement comprising a plurality of statement entities;

generating an original confidence estimate (CE) for a correctness of the original statement;

generating a replacement statement by;

selecting a suspect statement entity in the plurality of statement entities;

selecting a replacement entity for the suspect statement entity; and

replacing the suspect statement entity with the replacement entity in the original statement to generate a replacement statement;

generating a replacement CE for a correctness of the replacement statement;

sending the replacement statement to the user when the replacement CE exceeds the original CE by a tuning factor; and

repeating generating the replacement statement and generating the replacement CE when the replacement CE does not exceed the original CE by the tuning factor.

2. The computer-implemented method of claim 1, wherein generating the replacement statement and generating the replacement CE are repeated no more than a predetermined number of times.

3. The computer-implemented method of claim 1, wherein the tuning factor is set by the user.

4. The computer-implemented method of claim 1, wherein the original CE and the replacement CE are generated using a yes/no pipeline.

5. The computer-implemented method of claim 1, wherein selecting the suspect statement entity comprises:

determining connection strengths between each pair of statement entities in the plurality of statement entities using a knowledge graph; and

selecting as the suspect statement entity a statement entity in the plurality of statement entities having a lowest connection strength with other statement entities in the plurality of statement entities.

6. The computer-implemented method of claim 5, wherein selecting the replacement entity for the suspect statement entity comprises:

determining a type of the suspect statement entity;

finding one or more candidate entities having the same type as the suspect statement entity using the knowledge graph;

determining connection strengths between each of the one or more candidate entities and the statement entities other than the suspect statement entity in the plurality of statement entities using a knowledge graph; and

selecting as the replacement entity the candidate entity having a highest connection strength with the statement entities other than the suspect statement entity in the plurality of statement entities.

7. The computer-implemented method of claim 6, wherein determining the type of the suspect statement entity comprises:

determining a plurality of candidate types for the suspect statement entity and a corresponding type size for each of the plurality of candidate types; and

selecting a candidate type having the smallest type size as the type of the suspect statement entity.

8. A cognitive assistant (CA) system configured to improve an accuracy of a statement, the CA system comprising memory for storing instructions, and a processor configured to execute the instructions to:

receive from a user an original statement comprising a plurality of statement entities;

generate an original confidence estimate (CE) for a correctness of the original statement;

generate a replacement statement by;

selecting a suspect statement entity in the plurality of statement entities;

selecting a replacement entity for the suspect statement entity;

replacing the suspect statement entity with the replacement entity in the original statement to generate the replacement statement;

generate a replacement CE for a correctness of the replacement statement;

send the replacement statement to the user when the replacement CE exceeds the original CE by a tuning factor; and

repeat generating the replacement statement and generating the replacement CE when the replacement CE does not exceed the original CE by the tuning factor.

9. The CA system of claim 8, wherein the processor is configured to execute the instructions to repeat generating the replacement statement and generating the replacement CE no more than a predetermined number of times.

10. The CA system of claim 8, wherein the tuning factor is set by the user.

11. The CA system of claim 8, wherein the processor is configured to execute the instructions to:

send the original statement to a yes/no pipeline to generate the original CE; and

send the replacement statement to the yes/no pipeline to generate the replacement CE.

12. The CA system of claim 8, wherein the processor is configured to execute the instructions to:

send the plurality of statement entities to a knowledge graph to determine connection strengths between each pair of statement entities in the plurality of statement entities; and

select as the suspect statement entity a statement entity in the plurality of statement entities having a lowest connection strength with other statement entities in the plurality of statement entities.

13. The CA system of claim 12, wherein the processor is configured to execute the instructions to:

determine a type of the suspect statement entity;

send the type of the suspect statement entity to the knowledge graph to find one or more candidate entities having the same type as the suspect statement entity;

for each of the one or more candidate entities;

substitute the candidate entity for the suspect entity in the plurality of statement entities to create a candidate entities list; and

send the candidate entities list to the knowledge graph to determine connection strengths between each pair of statement entities in the candidate entities list; and

select as the replacement entity the candidate entity having a highest connection strength with the statement entities in its candidate entities list.

14. The CA system of claim 13, wherein the processor is configured to execute the instructions to:

determine a plurality of candidate types for the suspect statement entity and a corresponding type size for each of the plurality of candidate types; and

selecting a candidate type having the smallest type size as the type of the suspect statement entity.

15. A computer program product for improving an accuracy of a statement, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:

receive from a user an original statement comprising a plurality of statement entities;

generate an original confidence estimate (CE) for a correctness of the original statement;

generate a replacement statement by;

selecting a suspect statement entity in the plurality of statement entities;

selecting a replacement entity for the suspect statement entity;

replacing the suspect statement entity with the replacement entity in the original statement to generate the replacement statement;

generate a replacement CE for a correctness of the replacement statement;

send the replacement statement to the user when the replacement CE exceeds the original CE by a tuning factor; and

repeat generating the replacement statement and generating the replacement CE when the replacement CE does not exceed the original CE by the tuning factor.

16. The computer program product of claim 15, wherein the tuning factor is set by the user.

17. The computer program product of claim 15, the program instructions executable by the processor to cause the processor to:

send the original statement to a yes/no pipeline to generate the original CE; and

send the replacement statement to the yes/no pipeline to generate the replacement CE.

18. The computer program product of claim 17, the program instructions executable by the processor to cause the processor to:

send the plurality of statement entities to a knowledge graph to determine connection strengths between each pair of statement entities in the plurality of statement entities; and

select as the suspect statement entity a statement entity in the plurality of statement entities having a lowest connection strength with other statement entities in the plurality of statement entities.

19. The computer program product of claim 18, the program instructions executable by the processor to cause the processor to:

determine a type of the suspect statement entity;

send the type of the suspect statement entity to the knowledge graph to find one or more candidate entities having the same type as the suspect statement entity;

for each of the one or more candidate entities;

substitute the candidate entity for the suspect entity in the plurality of statement entities to create a candidate entities list; and

send the candidate entities list to the knowledge graph to determine connection strengths between each pair of statement entities in the candidate entities list; and

select as the replacement entity the candidate entity having a highest connection strength with the statement entities in its candidate entities list.

20. The computer program product of claim 19, the program instructions executable by the processor to cause the processor to:

determine a plurality of candidate types for the suspect statement entity and a corresponding type size for each of the plurality of candidate types; and

selecting a candidate type having the smallest type size as the type of the suspect statement entity.