Natural language question answering systems

Patent metadata

Field	Value
Publication number	US-12189691-B2
Application number	US-202318459887-A
Country	US
Kind code	B2
Filing date	Sep 1, 2023
Priority date	Mar 2, 2018
Publication date	Jan 7, 2025
Grant date	Jan 7, 2025

Abstract

A current set of context features for a database query that is associated with a string is identified. The database query includes a sequence of tokens of a database syntax, and the current set of context features includes words from the string and tokens from the database query. An inference record is selected from an inference store based on a comparison of the current set of context features to context features of inference records in the inference store. The database query is modified using a resolution of the inference record to obtain an inferred database query. The resolution includes one or more tokens of the database syntax. A search of a database is invoked using a query based on the inferred database query to obtain search results.

First claim

What is claimed is:

1. A method, comprising:
identifying a current set of context features for a database query that is associated with a string, wherein the database query includes a sequence of tokens of a database syntax, and the current set of context features includes words from the string and tokens from the database query;
selecting an inference record from an inference store based on a comparison of the current set of context features to context features of inference records in the inference store,
wherein the inference record includes a set of context features, a corresponding resolution, and an inference type selected from a set comprising superlative disambiguation and date disambiguation, and
wherein the inference record is learned based on user selections and feedback during query formulations;

modifying the database query using a resolution of the inference record to obtain an inferred database query, wherein the resolution includes one or more tokens of the database syntax; and
invoking a search of a database using a query based on the inferred database query to obtain search results.

2. The method of claim 1 , wherein the inference record comprises a confidence score and the inference record is selected based on the confidence score.

3. The method of claim 1 , wherein the current set of context features comprises part-of-speech tags for words of the string.

4. The method of claim 3 , further comprising:
comparing the part-of-speech tags of the current set of context features to part-of-speech tags of inference records in the inference store.

5. The method of claim 4 , wherein the inference record is selected based on a match of part-of-speech tags of the inference record to the part-of-speech tags of the current set of context features.

6. The method of claim 1 , further comprising:
receiving, via a user interface, an indication that the string matches a candidate database query;
responsive to the indication, determining a pattern based on the string and the candidate database query, wherein the pattern includes a collection of token constraints, a rewrite rule that maps one or more input tokens to a sequence of output tokens, and a ranking score adjustment that can be applied to a ranking score for the candidate database query; and
storing the pattern.

7. The method of claim 6 , wherein the indication is based on a user interaction with a like icon in the user interface while the user interface includes representations of the string and the candidate database query.

8. A system, comprising:
a memory; and
a processor, the processor configured to execute instructions stored in the memory to:
identify a current set of context features for a database query that is associated with a string, wherein the database query includes a sequence of tokens of a database syntax, and the current set of context features includes words from the string and tokens from the database query;
select an inference record from an inference store based on a comparison of the current set of context features to context features of inference records in the inference store,
wherein the inference record includes a set of context features, a corresponding resolution, and an inference type selected from a set comprising superlative disambiguation and date disambiguation, and
wherein the inference record is learned based on user selections and feedback during query formulations;
modify the database query using a resolution of the inference record to obtain an inferred database query, wherein the resolution includes one or more tokens of the database syntax; and
invoke a search of a database using a query based on the inferred database query to obtain search results.

9. The system of claim 8 , wherein the inference record comprises a confidence score and the inference record is selected based on the confidence score.

10. The system of claim 8 , wherein the current set of context features comprises part-of-speech tags for words of the string.

11. The system of claim 10 , wherein the processor is further configured to execute instructions stored in the memory to:
compare the part-of-speech tags of the current set of context features to part-of-speech tags of inference records in the inference store.

12. The system of claim 11 , wherein the inference record is selected based on a match of part-of-speech tags of the inference record to the part-of-speech tags of the current set of context features.

13. The system of claim 8 , wherein the processor is further configured to execute instructions stored in the memory to:
receive, via a user interface, an indication that the string matches a candidate database query;
responsive to the indication, determine a pattern based on the string and the candidate database query, wherein the pattern includes a collection of token constraints, a rewrite rule that maps one or more input tokens to a sequence of output tokens, and a ranking score adjustment that can be applied to a ranking score for the candidate database query; and
store the pattern.

14. The system of claim 13 , wherein the indication is based on a user interaction with a like icon in the user interface while the user interface includes representations of the string and the candidate database query.

15. A non-transitory computer readable medium storing instructions operable to cause one or more processors to perform operations comprising:
identifying a current set of context features for a database query that is associated with a string, wherein the database query includes a sequence of tokens of a database syntax, and the current set of context features includes words from the string and tokens from the database query;
selecting an inference record from an inference store based on a comparison of the current set of context features to context features of inference records in the inference store,
wherein the inference record includes a set of context features, a corresponding resolution, and an inference type selected from a set comprising superlative disambiguation and date disambiguation, and
wherein the inference record is learned based on user selections and feedback during query formulations;

modifying the database query using a resolution of the inference record to obtain an inferred database query, wherein the resolution includes one or more tokens of the database syntax; and
invoking a search of a database using a query based on the inferred database query to obtain search results.

16. The non-transitory computer readable medium of claim 15 , wherein the inference record comprises a confidence score and the inference record is selected based on the confidence score.

17. The non-transitory computer readable medium of claim 15 , wherein the current set of context features comprises part-of-speech tags for words of the string.

18. The non-transitory computer readable medium of claim 17 , wherein the operations further comprise:
comparing the part-of-speech tags of the current set of context features to part-of-speech tags of inference records in the inference store.

19. The non-transitory computer readable medium of claim 18 , wherein the inference record is selected based on a match of part-of-speech tags of the inference record to the part-of-speech tags of the current set of context features.

20. The non-transitory computer readable medium of claim 15 , wherein the operations further comprise:
receiving, via a user interface, an indication that the string matches a candidate database query;
re

Assignees

• Thoughtspot Inc

Inventors

• Prakash Amit
• Tandon Ravi
• Balakavi Manikanta Venkata Rahul
• Piratla Pavan Ram
• Shubham Ashish
• Canada Alonzo
• Kothari Rakesh
• Apte Maneesh
• Singhal Amitabh
• Viswanathan Aditya
• Singh Ajeet

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F16/243 Primary

  Natural language query formulation · CPC title

• G06F16/24534

  Query rewriting; Transformation · CPC title

• G06N5/04

  Inference or reasoning models · CPC title

• G06F16/90335 Primary

  Query processing · CPC title