Back to results
US 11,887,585 B2Granted
Global re-ranker
AssigneeApple Inc.
InventorsSrinivas Chappidi, Arash Dawoodi
Life of this patent
Estimating the term…
Abstract
Systems and processes for operating an intelligent automated assistant are provided. An example method includes, at an electronic device having one or more processors and memory: receiving a natural language speech input; determining, based on the natural language speech input, a plurality of candidate intents; obtaining contextual data associated with the user device; ranking, based on the contextual data, the plurality of candidate intents using a machine learning model, wherein the machine learning model is pre-trained at least partially on the user device; determining a user intent based on the ranked candidate intents; and performing a task corresponding to the determined user intent.
Claims63
A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by one or more processors of a first electronic user device, cause the first electronic user device to:
- receive a natural language speech input;
- determine, based on the natural language speech input, a plurality of candidate intents corresponding to the natural language speech input;
- receive a ranking of the plurality of candidate intents corresponding to the natural language speech input, wherein the ranking of the plurality of candidate intents is a ranking based on a plurality of natural-language processing confidence scores associated with the natural language speech input;
- after receiving the ranking of the plurality of candidate intents corresponding to the natural language speech input, obtain contextual data associated with the first electronic user device;
- adjust, based on the contextual data, the ranking of the plurality of candidate intents corresponding to the natural language speech input using a machine learning model to obtain an adjusted ranking of the plurality of candidate intents corresponding to the natural language speech input, wherein the machine learning model is pre-trained at least partially on a server with training data associated with users other than the user of the first electronic device and then subsequently trained on the first electronic user device with user specific training data without attempting to provide the machine learning model trained with the user specific training data, parameters of the machine learning model, and the user specific training data to the server;
- determine a user intent based on the adjusted ranking of the plurality of candidate intents corresponding to the natural language speech input;
- and
- perform a task corresponding to the determined user intent.
5 more elements
A first electronic user device comprising:
- one or more processors;
- a memory;
- and
- one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for:
- receiving a natural language speech input;
- determining, based on the natural language speech input, a plurality of candidate intents corresponding to the natural language speech input;
- receiving a ranking of the plurality of candidate intents corresponding to the natural language speech input, wherein the ranking of the plurality of candidate intents is a ranking based on a plurality of natural-language processing confidence scores associated with the natural language speech input;
- after receiving the ranking of the plurality of candidate intents corresponding to the natural language speech input, obtaining contextual data associated with the first electronic user device;
- adjusting, based on the contextual data, the ranking of the plurality of candidate intents corresponding to the natural language speech input using a machine learning model to obtain an adjusted ranking of the plurality of candidate intents corresponding to the natural language speech input, wherein the machine learning model is pre-trained at least partially on a server with training data associated with users other than a user of the first electronic user device and then subsequently trained on the first electronic user device with user specific training data without attempting to provide the machine learning model trained with the user specific training data, parameters of the machine learning model, and the user specific training data to the server;
- determining a user intent based on the adjusted ranking of the plurality of candidate intents corresponding to the natural language speech input;
- and
- performing a task corresponding to the determined user intent.
9 more elements
A method of determining a user intent, comprising:
- at a first electronic user device with one or more processors and memory:
- receiving a natural language speech input;
- determining, based on the natural language speech input, a plurality of candidate intents corresponding to the natural language speech input;
- receiving a ranking of the plurality of candidate intents corresponding to the natural language speech input, wherein the ranking of the plurality of candidate intents is a ranking based on a plurality of natural-language processing confidence scores associated with the natural language speech input,
- after receiving the ranking of the plurality of candidate intents corresponding to the natural language speech input, obtaining contextual data associated with the first electronic user device;
- adjusting, based on the contextual data, the ranking of the plurality of candidate intents corresponding to the natural language speech input using a machine learning model to obtain an adjusted ranking of the plurality of candidate intents corresponding to the natural language speech input, wherein the machine learning model is pre-trained at least partially on a server with training data associated with users other than a user of the first electronic user device and then subsequently trained on the first electronic user device with user specific training data without attempting to provide the machine learning model trained with the user specific training data, parameters of the machine learning model, and the user specific training data to the server;
- determining a user intent based on the adjusted ranking of the plurality of candidate intents corresponding to the natural language speech input;
- and
- performing a task corresponding to the determined user intent.
6 more elements
Dependencies are inferred from the claim text.