US 12,586,578 B1Granted

Natural language processing

AssigneeAmazon Technologies, Inc.

InventorsYing Shi, Joe Pemberton, Mariusz Momotko, Paul F. D. Ferraro, Andrew Smith, Melanie C B Gens

Life of this patent

Estimating the term…

Loading patent…

Abstract

Techniques for generating an executable API call for an LLM-generated request, where the executable API call is usable to cause a component to generate a potential response to a user input, are described. In some embodiments, the system receives a user input and uses a language model to generate a request for a component to provide a potential response to the user input. The system uses the request, an API description corresponding to the component, and other information not available to the language model during processing to generate an executable API call corresponding to the request. The system can execute the executable API calls (in a system-determined order or concurrently) to cause the corresponding components to generate potential responses to the user input.

Claims20

Dependent claims

A computer-implemented method comprising:

receiving first input audio data representing a first spoken input;
performing automatic speech recognition (ASR) processing using the first input audio data to generate a first transcript of the first spoken input;
processing, by a first language model, the first transcript and first context data to generate a first request for a first component to generate first potential response data;
receiving, from the first language model, the first request;
receiving second context data associated with the first spoken input, wherein the second context data is different from the first context data processed by the first language model;
determining, using the first request, a first component description associated with the first component, wherein the first component description includes one or more functions performable by the first component and a parameter usable by the first component to perform the one or more functions;
determining a value for the parameter using the second context data;
using the first transcript, the first request, the first component description, and the second context data, generating a first application programming interface (API) call requesting the first component to generate the first potential response data, wherein the first API call includes the value usable by the first component to generate the first potential response data;
using the first API call, causing the first component to generate the first potential response data;
generating, based at least in part on the first potential response data, first output data responsive to the first spoken input;
and
causing presentation of the first output data.

9 more elements

A computer-implemented method comprising:

receiving first input data;
determining, using first data and the first input data, a first request for a first component to generate first potential response data;
receiving second data corresponding to the first input data, the second data being different than the first data;
based at least in part on the first input data, the first request and the second data, determining a first application programming interface (API) call executable by the first component, the first API call including a first parameter usable by the first component to generate the first potential response data, wherein the first parameter corresponds to the second data;
based at least in part on the first API call, receiving, from the first component, the first potential response data;
generating, based at least in part on the first potential response data, first output data responsive to the first input data;
and
causing presentation of the first output data.

5 more elements

A computing system comprising:

at least one processor;
and
at least one memory including instruction that, when executed by the at least one processor, cause the computing system to:
receive first input data;
determine, using first data and the first input data, a first request for a first component to generate first potential response data;
receive second data corresponding to the first input data, the second data being different than the first data;
based at least in part on the first input data, the first request and the second data determine a first application programming interface (API) call executable by the first component, the first API call including a first parameter usable by the first component to generate the first potential response data, wherein the first parameter corresponds to the second data;
based at least in part on the first API call, receive, from the first component, the first potential response data;
generate, based at least in part on the first potential response data, first output data responsive to the first input data;
and
cause presentation of the first output data.

8 more elements

Dependencies are inferred from the claim text.