DYNAMIC CHOICE OF DATA SOURCES IN NATURAL LANGUAGE QUERY PROCESSING

• US 20180121508A1
• Filed: 11/03/2016
• Published: 05/03/2018
• Est. Priority Date: 11/03/2016
• Status: Active Grant

• Current Assignee: Soundhound AI IP Holding LLC (SoundHound AI, Inc. (f/k/a Archimedes Tech SPAC Partners Co.)), Soundhound AI IP LLC (SoundHound AI, Inc. (f/k/a Archimedes Tech SPAC Partners Co.))
• Original Assignee: SoundHound, Inc. (SoundHound AI, Inc. (f/k/a Archimedes Tech SPAC Partners Co.))
• Inventors: Halstvedt, Scott
• Granted Patent: US 10,585,891 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/24545 Selectivity estimation or determination
• CPC Class Codes: G06F 16/3344 using natural language analysis

Abstract

A virtual assistant receives natural language interpretation hypotheses for user queries, determines entities and attributes from the interpretations, and requests data from appropriate data sources. A cost function estimates the cost of each data source request. Cost functions include factors such as contract pricing, access latency, and data quality. Based on the estimated cost, the virtual assistant sends requests to a plurality of data sources, each of which might be able to provide data necessary to answer the user query. By including user credits in the cost function, the virtual assistant provides better quality of results and answer latency for paying users. The virtual assistant minimizes latency by answering using data from the first responding data source or provides a latency guarantee by answering with the most accurate data received by a deadline. The virtual assistant measures data source response latency and caches responses for expensive requests.

Claims

1. A method of processing interpretations of natural language queries, the method comprising:

determining, by a computer system, a hypothesized interpretation of a natural language query, the answer to which depends on a value of particular data;

determining, by the computer system, a plurality of data sources that might be able to provide the value;

estimating, by the computer system, for at least one data source of the plurality of data sources, a cost for getting the value; and

identifying, by the computer system, a lowest cost data source of the plurality of data sources.

2. The method of claim 1 wherein determining the plurality of electronic data sources is a function of a type of the particular data.

3. The method of claim 1 wherein determining the plurality of electronic data sources is a function of dynamic availability of the at least one data source.

4. The method of claim 1 wherein determining the plurality of electronic data sources is a function of the domain of discourse.

5. The method of claim 1 wherein determining the plurality of data sources is a function of a history of recent queries.

6. The method of claim 1, wherein estimating, for the at least one data source, the cost for getting the value depends on the contracted monetary consideration for accessing the at least one data source.

7. The method of claim 1, wherein estimating, for the at least one data source, the cost for getting the value depends on an expected latency for accessing the at least one data source.

8. The method of claim 1, wherein estimating, for the at least one data source, the cost for getting the value depends on a quality of the at least one data source.

9. The method of claim 1, wherein estimating, for the at least one data source, the cost for getting the value comprises executing a cost function.

10. The method of claim 9, wherein the cost function depends on an attribute of a second data source.

11. The method of claim 9, further comprising receiving feedback, from a user, regarding a previous answer, wherein the cost function depends on the feedback.

12. The method of claim 9 further comprising measuring latency for the at least one data source;

and using the latency as an input to the cost function.

13. The method of claim 9 further comprising recording the availability of the at least one data source;

and using the availability as an input to the cost function.

14. The method of claim 1 further comprising:

requesting the value from the lowest cost data source;

receiving the value from the lowest cost data source; and

using the value in an answer to the query.

15. The method of claim 1 further comprising identifying a second data source.

16. The method of claim 15 further comprising:

requesting the value from the lowest cost data source;

requesting the value from the second data source;

receiving the value from either of the lowest cost data source and the second data source; and

using the value in an answer to the query.

17. The method of claim 15 further comprising:

requesting the value from the lowest cost data source;

requesting the value from the second data source;

receiving a first data value from the lowest cost data source;

receiving a second data value from the second data source; and

choosing a chosen data value from among the first data value and the second data value; and

using the chosen data value in an answer to the query.

18. The method of claim 17 further comprising validating the first data value to determine a confirmation, wherein the choosing is based on the confirmation.

19. The method of claim 17 further comprising comparing the first data value and the second data value to determine a difference, wherein the choosing is based on the difference.

20. The method of claim 15 further comprising:

requesting the value from the lowest cost data source;

requesting the value from the second data source;

receiving a first received data value from one of the first data source and the second data source;

responsive to receiving a second received data value from another of the first data source and the second data source within an allowable time period, choosing a chosen data value from among the first data value and the second data value; and

using the chosen data value in an answer to the query.

21. The method of claim 1 further comprising, receiving an alternative interpretation of the natural language query, the answer to which depends on the value of the particular data, wherein determining the plurality of electronic data sources includes identifying data sources appropriate to at least one of the hypothesized interpretation and the alternative interpretation.

22. The method of claim 1 further comprising:

requesting the value from the at least one data source;

receiving the value from the at least one data source;

storing the value in a cache; and

using the value from the cache in a future query answers.

23. The method of claim 1 wherein the method is performed on a many-core chip optimized for the minimum power consumption needed for artificial intelligence processing tasks.

24. The method of claim 1 further comprising:

requesting the value from the at least one data source;

receiving the value from the at least one data source; and

refining the hypothesized interpretation based on the value from the at least one data source.

25. At least one non-transitory computer readable medium comprising code that, if executed by at least one computer processor, would cause the at least one computer processor to:

determine a multiplicity of hypothesized interpretations of a natural language query, the answers to each of which depend on at least one data value;

determine that at least two of the hypothesized interpretations depend on one particular value;

estimate, for at least one data source of a plurality of data sources that can provide the particular value, a cost for getting the particular value; and

generate only one request for the particular data value to each of the plurality of data sources,

wherein the data sources in the plurality of data sources are selected according to a cost function based on the cost.

26. The at least one non-transitory computer readable medium of claim 25, wherein the code would cause a first computer processor to determine the multiplicity of hypothesized interpretations and a second computer processor to generate the only one request to each of the plurality of data sources.