Dynamic gazetteers for personalized entity recognition
AssigneeAmazon Technologies, Inc.
InventorsImre Attila Kiss, Arthur Richard Toth, Lambert Mathias
Life of this patent
Estimating the term…
Abstract
In speech processing systems personalization is added in the Natural Language Understanding (NLU) processor by incorporating external knowledge sources of user information to improve entity recognition performance of the speech processing system. Personalization in the NLU is effected by incorporating one or more dictionaries of entries, or gazetteers, with information personal to a respective user, that provide the user's information to permit disambiguation of semantic interpretation for input utterances to improve quality of speech processing results.
Claims21
1-22. (canceled)
A computer implemented method, comprising:
- obtaining a textual representation of a user request, the user request associated with user identification data;
- obtaining first user specific data associated with the user identification data;
- creating a first feature vector corresponding to the textual representation, the first feature vector including a first feature value indicating that a first word of the textual representation is represented in the first user specific data;
- determining, based at least in part on the first feature vector, first classification data indicating a first classification associated with one or more words included in the textual representation;
- using the textual representation and the first classification data to determine intent data;
- and
- sending the intent data to a component to execute a command associated with the intent data.
A computer implemented method, comprising:
- obtaining a textual representation of a user request, the user request associated with a device;
- obtaining first device specific data associated with the device;
- creating a feature vector corresponding to the textual representation, the feature vector including a first feature value indicating that a first word of the textual representation is represented in the first device specific data;
- determining, based at least in part on the feature vector, first classification data indicating a first classification associated with one or more words included in the textual representation;
- using the textual representation and the first classification data to determine intent data;
- and
- sending the intent data to a component to execute a command associated with the intent data.
A computer implemented method, comprising:
- obtaining a textual representation of a user request;
- obtaining location data associated with the user request;
- creating a first feature vector corresponding to the textual representation, the first feature vector including a first feature value indicating that a first word of the textual representation is represented in the location data;
- determining, based at least in part on the first feature vector, first classification data indicating a first classification associated with one or more words included in the textual representation;
- using the textual representation and the first classification data to determine intent data;
- and
- sending the intent data to a component to execute a command associated with the intent data.
Dependencies are inferred from the claim text.