Context tag integration with named entity recognition models
Techniques are provided for using context tags in named-entity recognition (NER) models. In one particular aspect, a method is provided that includes receiving an utterance, generating embeddings for words of the utterance, generating a regular expression and gazetteer feature vector for the utterance, generating a context tag distribution feature vector for the utterance, concatenating or interpolating the embeddings with the regular expression and gazetteer feature vector and the context tag distribution feature vector to generate a set of feature vectors, generating an encoded form of the utterance based on the set of feature vectors, generating log-probabilities based on the encoded form of the utterance, and identifying one or more constraints for the utterance.
1. A method comprising:
accessing an utterance comprising a plurality of tokens;
associating a token of the plurality of tokens with a plurality of candidate entities;
determining a plurality of confidence scores for the plurality of candidate entities, wherein a confidence score is determined for each candidate entity of the plurality of candidate entities;
generating a context tag distribution vector for the utterance based on the plurality of confidence scores;
generating a plurality of embeddings for the plurality of tokens;
identifying one or more constraints for the utterance based on the context tag distribution vector and the plurality of embeddings; and
identifying an intent associated with the utterance based on at least one constraint of the one or more constraints.
2. The method of claim 1 , wherein the plurality of embeddings for the plurality of tokens are generated using a transformer-based machine learning model.
3. The method of claim 1 , further comprising:
generating a feature vector for the utterance based on at least one of a regular expression pattern and a gazetteer.
4. The method of claim 3 , further comprising:
processing the context tag distribution vector, the plurality of embeddings, and the feature vector to generate a first set of feature vectors.
5. The method of claim 1 , further comprising:
generating an encoded form of the utterance based on the context tag distribution vector and the plurality of embeddings.
6. The method of claim 5 , further comprising:
generating a plurality of probabilities for the plurality of candidate entities based on the encoded form of the utterance.
7. The method of claim 1 , wherein the plurality of confidence scores for the plurality of candidate entities are determined based on a dialog comprised of a plurality of utterances.
8. A system comprising:
one or more processors; and
a memory storing instructions which, when executed by the one or more processors, cause the system to perform operations comprising:
accessing an utterance comprising a plurality of tokens;
associating a token of the plurality of tokens with a plurality of candidate entities;
determining a plurality of confidence scores for the plurality of candidate entities, wherein a confidence score is determined for each candidate entity of the plurality of candidate entities;
generating a context tag distribution vector for the utterance based on the plurality of confidence scores;
generating a plurality of embeddings for the plurality of tokens;
identifying one or more constraints for the utterance based on the context tag distribution vector and the plurality of embeddings; and
identifying an intent associated with the utterance based on at least one constraint of the one or more constraints.
9. The system of claim 8 , wherein the plurality of embeddings for the plurality of tokens are generated using a transformer-based machine learning model.
10. The system of claim 8 , further comprising:
generating a feature vector for the utterance based on at least one of a regular expression pattern and a gazetteer.
11. The system of claim 10 , further comprising:
processing the context tag distribution vector, the plurality of embeddings, and the feature vector to generate a first set of feature vectors.
12. The system of claim 8 , further comprising:
generating an encoded form of the utterance based on the context tag distribution vector and the plurality of embeddings.
13. The system of claim 12 , further comprising:
generating a plurality of probabilities for the plurality of candidate entities based on the encoded form of the utterance.
14. The system of claim 8 , wherein the plurality of confidence scores for the plurality of candidate entities are determined based on a dialog comprised of a plurality a plurality of utterances.
15. A non-transitory computer-readable medium storing instructions which, when executed by one or more processors, cause a system to perform operations comprising:
accessing an utterance comprising a plurality of tokens;
associating a token of the plurality of tokens with a plurality of candidate entities;
determining a plurality of confidence scores for the plurality of candidate entities, wherein a confidence score is determined for each candidate entity of the plurality of candidate entities;
generating a context tag distribution vector for the utterance based on the plurality of confidence scores;
generating a plurality of embeddings for the plurality of tokens;
identifying one or more constraints for the utterance based on the context tag distribution vector and the plurality of embeddings; and
identifying an intent associated with the utterance based on at least one constraint of the one or more constraints.
16. The non-transitory computer-readable medium of claim 15 , further comprising:
generating a feature vector for the utterance based on at least one of a regular expression pattern and a gazetteer.
17. The non-transitory computer-readable medium of claim 16 , further comprising:
processing the context tag distribution vector, the plurality of embeddings, and the feature vector to generate a first set of feature vectors.
18. The non-transitory computer-readable medium of claim 15 , further comprising:
generating an encoded form of the utterance based on the context tag distribution vector and the plurality of embeddings.
19. The non-transitory computer-readable medium of claim 18 , further comprising:
generating a plurality of probabilities for the plurality of candidate entities based on the encoded form of the utterance.
20. The non-transitory computer-readable medium of claim 15 , wherein the plurality of confidence scores for the plurality of candidate entities are determined based on a dialog comprised of a plurality a plurality of utterances.