Optimizing retrieval of data related to temporal based queries

Patent metadata

Field	Value
Publication number	US-10198501-B2
Application number	US-201615278151-A
Country	US
Kind code	B2
Filing date	Sep 28, 2016
Priority date	Sep 28, 2016
Publication date	Feb 5, 2019
Grant date	Feb 5, 2019

Abstract

A computer-implemented method generates a candidate answer triple for use in retrieving information used to answer a question. One or more processors parse a question to identify a lexical answer type for the question, a question action for the question, and a question timestamp for the question to make up a question triple. One or more processors retrieve multiple candidate passages for answering the question, and parse each of the multiple candidate passages to identify a candidate entity, a candidate action, and a candidate timestamp from each of the multiple candidate passages to generate a candidate answer triple. One or more processors compare the question triple to the candidate answer triple and establish a match score for each candidate answer triple, which is used in retrieving information used to answer the question.

First claim

What is claimed is:

1. A computer-implemented method comprising:
parsing, by one or more processors, a question to identify a lexical answer type for the question, a question action for the question, and a question timestamp for the question, wherein the lexical answer type indicates what entity type is being asked for by the question, wherein the question action describes an action referenced in the question that was performed by an entity of the entity type, wherein the question timestamp defines a time range during which the action referenced in the question was performed by the entity, and wherein the lexical answer type, the question action, and the question timestamp make up a question triple;
retrieving, by one or more processors, multiple candidate passages for answering the question;
parsing, by one or more processors, each of the multiple candidate passages to identify a candidate entity, a candidate action, and a candidate timestamp from each of the multiple candidate passages;
generating, by one or more processors, a candidate answer triple from the candidate entity, the candidate action, and the candidate timestamp identified from each of the multiple candidate passages;
comparing, by one or more processors, the question triple to the candidate answer triple generated from each of the multiple candidate passages, wherein the entity type in the question triple is compared to the candidate entity in the candidate answer triple, wherein the question action in the question triple is compared to the candidate action in the candidate answer triple, and wherein the question timestamp in the question triple is compared to the candidate timestamp in the candidate answer triple;
establishing, by one or more processors, a match score for each candidate answer triple being compared to the question triple, wherein the match score is based on how closely the question triple matches a particular candidate answer triple;
tagging, by one or more processors, the match score to each corresponding candidate answer triple generated from the multiple candidate passages; and
associating, by one or more processors, the match score for each corresponding candidate answer triple with a corresponding candidate passage that generated the corresponding candidate answer triple for use in retrieving information used to answer the question.

2. The computer-implemented method of claim 1 , wherein the match score for each candidate answer triple is established by semantically matching each candidate answer triple to the question triple.

3. The computer-implemented method of claim 1 , wherein the match score for each candidate answer triple is established by stem word matching each candidate answer triple to the question triple.

4. The computer-implemented method of claim 1 , wherein the match score for each candidate answer triple is established by synonym matching each candidate answer triple to the question triple.

5. The computer-implemented method of claim 1 , wherein the lexical answer type defines an event that the question is asking about.

6. The computer-implemented method of claim 1 , wherein the lexical answer type defines a person that the question is asking about.

7. The computer-implemented method of claim 1 , wherein the lexical answer type defines a place that the question is asking about.

8. A computer program product comprising one or more computer readable storage mediums, and program instructions stored on at least one of the one or more storage mediums, the stored program instructions comprising:
program instructions to parse a question to identify a lexical answer type for the question, a question action for the question, and a question timestamp for the question, wherein the lexical answer type indicates what entity type is being asked for by the question, wherein the question action describes an action referenced in the question that was performed by an entity of the entity type, wherein the question timestamp defines a time range during which the action referenced in the question was performed by the entity, and wherein the lexical answer type, the question action, and the question timestamp make up a question triple;
program instructions to retrieve multiple candidate passages for answering the question;
program instructions to parse each of the multiple candidate passages to identify a candidate entity, a candidate action, and a candidate timestamp from each of the multiple candidate passages;
program instructions to generate a candidate answer triple from the candidate entity, the candidate action, and the candidate timestamp identified from each of the multiple candidate passages;
program instructions to compare the question triple to the candidate answer triple generated from each of the multiple candidate passages, wherein the entity type in the question triple is compared to the candidate entity in the candidate answer triple, wherein the question action in the question triple is compared to the candidate action in the candidate answer triple, and wherein the question timestamp in the question triple is compared to the candidate timestamp in the candidate answer triple;
program instructions to establish a match score for each candidate answer triple being compared to the question triple, wherein the match score is based on how closely the question triple matches a particular candidate answer triple;
program instructions to tag the match score to each corresponding candidate answer triple generated from the multiple candidate passages; and
program instructions to associate the match score for each corresponding candidate answer triple with a corresponding candidate passage that generated the corresponding candidate answer triple for use in retrieving information used to answer the question.

9. The computer program product of claim 8 , wherein the match score for each candidate answer triple is established by semantically matching each candidate answer triple to the question triple.

10. The computer program product of claim 8 , wherein the match score for each candidate answer triple is established by stem word matching each candidate answer triple to the question triple.

11. The computer program product of claim 8 , wherein the match score for each candidate answer triple is established by synonym matching each candidate answer triple to the question triple.

12. The computer program product of claim 8 , wherein the lexical answer type defines an event that the question is asking about.

13. The computer program product of claim 8 , wherein the lexical answer type defines a person that the question is asking about.

14. The computer program product of claim 8 , wherein the lexical answer type defines a place that the question is asking about.

15. The computer program product of claim 8 , wherein the program instructions are provided as a service in a cloud environment.

16. A computer system comprising one or more processors, one or more computer readable memories, and one or more computer readable storage mediums, and program instructions stored on at least one of the one or more storage mediums for execution by at least one of the one or more processors via at least one of the one or more memories, the stored program instructions comprising:
program instructions to parse a question to identify a lexical answer type for the question, a question action for the question, and a question timestamp for the question, wherein the lexical answer type indicates what entity type is being asked for by the question, wherein the question action describes an action referenced in the question that was

Assignees

• IBM

Inventors

• Croutwater Kyle L
• Verma Vikrant
• Zhang Le
• Zhang Zhe

Classifications

• G06F40/30

  Semantic analysis · CPC title

• G06F16/3334

  Selection or weighting of terms from queries, including natural language queries · CPC title

• G06F16/3329 Primary

  Natural language query formulation · CPC title

• G06F17/30663

  Physics · mapped topic

• G06F17/2785

  Physics · mapped topic