US 11,380,308 B1 · Natural language processing

Abstract

Devices and techniques are generally described for using user feedback to determine routing decisions in a speech processing system. In various examples, first data representing a first utterance may be received. Second data representing a first semantic interpretation of the first utterance may be determined. A first intent data processing application may be selected for processing the second data. Feedback data may be determined related to the first intent data processing application processing the second data. Third data representing a semantic interpretation of a second utterance may be received, wherein the first semantic interpretation is the same as the second semantic interpretation. A second intent data processing application may be determined for processing the third data based at least in part on the feedback data.

Claims20

Dependent claims

A method comprising:

receiving audio data representing a first spoken request;
determining, from the audio data, first intent data corresponding to a first skill, the first intent data representing a first semantic interpretation of the first spoken request and the first skill configured to generate output data from the first intent data;
determining, by a feedback prediction machine learning model, predicted user feedback data, the predicted user feedback representing feedback predicted for a user resulting from processing the first intent data by the first skill;
inputting the predicted user feedback data into a machine learning model of a selection component of a speech processing system;
determining, using the selection component, a first ranked list of a first plurality of skills to process the first intent data;
selecting the first skill from the first ranked list to process the first intent data;
sending the first intent data to the first skill;
and
generating the output data by the first skill using the first intent data.

6 more elements

A computing device, comprising:

at least one processor;
and
at least one non-transitory computer-readable memory storing instructions that, when executed by the at least one processor, are effective to program the at least one processor to:
receive first data representing a first natural language input;
determine second data representing a first semantic interpretation of the first natural language input;
determine a first application to process the second data;
generate, by a first machine learned model, third data related to processing the second data by the first application, the third data representing a prediction of user-generated feedback associated with the processing of the second data by the first application;
and
generate, by a first component, a group of applications able to process the second data based at least in part on the third data.

6 more elements

Dependencies are inferred from the claim text.

Natural language processing

Life of this patent

Natural language processing

Life of this patent

Abstract

Claims20