Interpreting queries according to preferences
The present invention extends to methods, systems, and computer program products for interpreting queries according to preferences. Multi-domain natural language understanding systems can support a variety of different types of clients. Queries can be received and interpreted across one or more domains. Preferred query interpretations can be identified and query responses provided based on any of: domain preferences, preferences indicated by an identifier, or (e.g., weighted) scores exceeding a threshold.
1. A method performed by a multi-domain natural language understanding (NLU) interpreting server comprising:
receiving a query and a domain preference from a client device, wherein the client device is one of a smart speaker, mobile phone, automobile, or vending machine, wherein the domain preference is stored as a weighted graph representing domain relationships and priorities;
interpreting the query according to a plurality of domains to produce a plurality of interpretations, wherein each interpretation is associated with a confidence score;
identifying a preferred interpretation from the plurality of interpretations by traversing the weighted graph to adjust the confidence scores of each domain based on the domain preference, said identification of the preferred interpretation increasing an accuracy of domain selection for ambiguous queries compared to a system without domain preferences; and
providing a query response for the preferred interpretation to the client device.
2. The method of claim 1 wherein providing a query response comprises invoking an action.
3. The method of claim 1 wherein providing a query response comprises providing a visual response or an audible response.
4. The method of claim 1 , wherein the domain preference includes an identifier.
5. The method of claim 4 wherein the identifier indicates a type of client device.
6. The method of claim 4 wherein the identifier indicates a make of automobile.
7. The method of claim 4 wherein the identifier indicates which application is running in the foreground of a client device.
8. The method of claim 1 , wherein providing the query response comprises invoking an action.
9. The method of claim 1 , wherein providing a query response comprises providing a visual response or an audible response.
10. A method performed by a multi-domain natural language understanding (NLU) interpreting server comprising:
receiving a query and an identifier from a client device, wherein the client device is one of a smart speaker, mobile phone, automobile, or vending machine;
referencing a numerical weight value indicated by the identifier;
interpreting the query according to a domain to produce an interpretation and a score;
adjusting the score by the numerical weight value to compute a weighted score;
comparing the weighted score to a determined threshold; and
providing a query response in dependence upon the weighted score exceeding the threshold to the client device.
11. The method of claim 10 wherein the threshold is a second weighted score for a second domain.
12. The method of claim 10 wherein the identifier indicates a type of client device.
13. The method of claim 10 wherein the identifier indicates a make of automobile.
14. The method of claim 10 wherein the identifier indicates which application is running in the foreground of a client device.
15. The method of claim 10 wherein providing the query response comprises invoking an action.
16. The method of claim 10 wherein providing a query response comprises providing a visual response or an audible response.