Interpreting nonstandard terms in language processing using text-based communications
First Claim
Patent Images
1. A system comprising:
- a computer-readable memory storing executable instructions; and
one or more processors in communication with the computer-readable memory, wherein the one or more processors are programmed by the executable instructions to at least;
receive first data regarding a first utterance of a first user of a plurality of users;
generate a transcription of the first utterance using an automatic speech recognition (“
ASR”
) subsystem, the transcription comprising a plurality of terms;
identify a first term of the plurality of terms as a named entity based at least partly on a part of speech of the first term and a co-occurrence in the first utterance of the first term and a second term, wherein the second term corresponds to an executable action, and wherein the executable action is to be executed based at least partly on the named entity;
determine, using a natural language understanding (“
NLU”
) subsystem, that the named entity is a nonstandard named entity based at least partly on a definition of the named entity not being resolved to a threshold degree of confidence;
determine that the first term is not included as an alias for a different term in a first set of aliases stored in a user data store, wherein the first set of aliases stored in the user data store is personalized to the first user;
after determining that the first term is not included in the first set of aliases personalized for the first user, determine that the first term is not included in a second set of aliases stored in a shared data store, wherein the second set of aliases stored in the shared data store is common to the plurality of users; and
after determining that the first term is not included in the first set of aliases personalized for the first user and is not included in the second set of aliases common to the plurality of users;
analyze text-based communications of the first user to identify an occurrence of a third term;
determine that the first term is an alias for the third term based at least partly on a context of the occurrence of the third term;
include the first term as the alias for the third term in the first set of aliases; and
execute the executable action in response to the utterance based at least partly on the first term.
1 Assignment
0 Petitions
Accused Products
Abstract
Features are disclosed for determining a definition or value of a nonstandard term. A user utterance may be processed into one or more candidate transcriptions. An interpretation of the utterance can be generated from the transcriptions. If the transcription includes a word, phrase, or term that is not recognized or is used in a nonstandard way, one or more data stores may be queried regarding the proper value or definition of the term. If a definition or value is not available in the data stores, the user may be prompted to provide one. The user-supplied definition can be saved for future use, and may be used as a general definition of the term for other users.
20 Citations
32 Claims
-
1. A system comprising:
-
a computer-readable memory storing executable instructions; and one or more processors in communication with the computer-readable memory, wherein the one or more processors are programmed by the executable instructions to at least; receive first data regarding a first utterance of a first user of a plurality of users; generate a transcription of the first utterance using an automatic speech recognition (“
ASR”
) subsystem, the transcription comprising a plurality of terms;identify a first term of the plurality of terms as a named entity based at least partly on a part of speech of the first term and a co-occurrence in the first utterance of the first term and a second term, wherein the second term corresponds to an executable action, and wherein the executable action is to be executed based at least partly on the named entity; determine, using a natural language understanding (“
NLU”
) subsystem, that the named entity is a nonstandard named entity based at least partly on a definition of the named entity not being resolved to a threshold degree of confidence;determine that the first term is not included as an alias for a different term in a first set of aliases stored in a user data store, wherein the first set of aliases stored in the user data store is personalized to the first user; after determining that the first term is not included in the first set of aliases personalized for the first user, determine that the first term is not included in a second set of aliases stored in a shared data store, wherein the second set of aliases stored in the shared data store is common to the plurality of users; and after determining that the first term is not included in the first set of aliases personalized for the first user and is not included in the second set of aliases common to the plurality of users; analyze text-based communications of the first user to identify an occurrence of a third term; determine that the first term is an alias for the third term based at least partly on a context of the occurrence of the third term; include the first term as the alias for the third term in the first set of aliases; and execute the executable action in response to the utterance based at least partly on the first term. - View Dependent Claims (2, 3)
-
-
4. A computer-implemented method comprising:
under control of one or more computing devices configured with specific computer-executable instructions, receiving audio data representing a first utterance of a first user in a plurality of users; generating a transcription of the first utterance using automatic speech recognition and the audio data, wherein the transcription comprises a first plurality of terms; identifying a first term of the first plurality of terms as an intent that corresponds to an executable action; identifying a second term of the first plurality of terms as a named entity, wherein the executable action is to be executed based at least partly on the named entity, and wherein the second term is identified based at least partly on a part of speech of the second term and a co-occurrence of the second term with the first term in the transcription; determining that the second term is a nonstandard term based at least on a definition of the nonstandard term not being determined to a threshold degree of confidence; determining that the definition of the nonstandard term is not included in information personalized for the first user; after determining that the definition of the nonstandard term is not included in the information personalized for the first user, determining that the definition of the nonstandard term is not included in information common to the plurality of users; after determining that the definition of the nonstandard term is not included in the information common to the plurality of users; analyzing text-based communications of the first user for information regarding the nonstandard term; determining the definition of the nonstandard term, the definition based at least partly on a context of an occurrence of the nonstandard term in the text-based communications; storing definition data regarding the definition of the nonstandard term in a data store; receiving a representation of a second utterance of the first user, the representation of the second utterance comprising a second plurality of terms; identifying the nonstandard term in the second plurality of terms; and executing the executable action in response to the second utterance based at least partly on the definition data. - View Dependent Claims (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24)
-
25. One or more non-transitory computer-readable media comprising executable code that, when executed, cause one or more computing devices to perform a process comprising:
-
receiving audio data representing a first utterance of a user of a plurality of users; generating a transcription of the first utterance using automatic speech recognition and the audio data, the transcription comprising a first plurality of terms; identifying a first term of the first plurality of terms as an intent that corresponds to an executable action; identifying a second term of the first plurality of terms as a named entity, wherein the executable action is to be executed based at least partly on the named entity, and wherein the second term is identified based at least partly on a part of speech of the second term and a co-occurrence of the second term with the first term in the transcription; determining that the second term is a nonstandard term based at least on a definition of the nonstandard term not being determined to a threshold degree of confidence; determining that the definition of the nonstandard term is not included in information personalized for the user; after determining that the definition of the nonstandard term is not included in the information personalized for the user; analyzing text-based communications of the user for information regarding the nonstandard term; determining the definition of the nonstandard term based at least partly on a context of an occurrence of the nonstandard term in the text-based communications; storing definition data regarding the definition of the nonstandard term in a data store; receiving a representation of a second utterance of the user, the second utterance comprising a second plurality of terms; identifying the nonstandard term in the second plurality of terms; and executing the executable action in response to the second utterance based at least partly on the definition data. - View Dependent Claims (26, 27, 28, 29, 30, 31, 32)
-
Specification