Dialogue act classification is the task of classifying an utterance with respect to the function it serves in a dialogue, i.e. the act the speaker is performing. This python library essentially does dialogue act classification on the Switchboard corpus.

The Switchboard-1 corpus is a telephone speech corpus, consisting of about 2,400 two-sided telephone conversation among 543 speakers with about 70 provided conversation topics. The dataset includes the audio files and the transcription files, as well as information about the speakers and the calls. A subset of the Switchboard-1 corpus consisting of 1155 conversations was used. The resulting tags include dialogue acts like statement-non-opinion, acknowledge, statement-opinion, agree/accept, etc.

Annotated example:

Speaker: A, Dialogue Act: Yes-No-Question, Utterance: So do you go to college right now?

The original dataset contained around 42 tags but here we brought them down to 38 by removing a few redundant and ad-hoc tags. The available tags:

We recommend Python 3.7 or higher, Tensorflow 2.0.0 or higher and Transformers v3.0.0 or higher.

Install the DialogTag with pip:

pip install -U DialogTag

Alternatively, you can also clone the latest version from the repository and install it directly from the source code:

pip install -e .

This quick example will show you how to use DialogTag in your code.

from dialog_tag import DialogTag

model = DialogTag('distilbert-base-uncased')

sentence = "I'll probably go to shopping today."
output = model.predict_tag(sentence)
print(output)
# output: 'Statement-non-opinion'

sentence = "Why are you asking me this question again and again?"
output = model.predict_tag(sentence)
print(output)
# output: 'Wh-Question'

Update: We now have bert-base-uncased model available!

Currently we have only distilbert-base-uncased available. We're planning to extend it to bert-base-uncased and roberta-base. Stay tuned for further updates!

This project is licensed under the MIT License - see the LICENSE.md file for details