Discourse-Wizard

Discourse-Wizard is to demonstrate the context in spoken language which comes from their sequential patterns. For example, a sentence is formed by a sequence of words, a conversation is formed by a sequence of utterances, and so on. Dialogue act represents a performative action of an utterance. There have been many approaches to model such a deep contextual-concept to analyse conversations. The neural approaches have been deployed and achieved state-of-the-art results. In this demonstration recurrent neural networks are used to model the context-based learning of dialogue acts.

Same utterance can be used in different contexts

not supported in your browser

As you can see in the small piece of conversation example, Utt2 and Utt4 are very same (in this case they are same “Yeah”), however, the dialogue acts are very different. As they come from their context utterances Utt1 and Utt3 respectively. If it appears after a ‘Yes-No Question’ dialogue act it is more likely to have ‘Yes-Answer’ rather than ‘Backchannel’ or any other dialogue act. The presented example is from the Switchboard Dialogue Act (SwDA) Corpus, which is annotated with 42 such dialogue acts.

Features:

  • Discourse/conversational analysis
  • Dialogue act recognition at utterance level
  • Context-based dialogue act recognition, preceding utterances as a context

Demonstration

Discourse-Wizard:- Dialogue Act Recognition Demo

News:

  • August 2018 - A preprint of our paper on Discourse-Wizard is now available; with title
    "Discourse-Wizard: Discovering Deep Discourse Structure in your Conversation with RNNs"
  • August 2018 - Code for the training and demonstraiton will be made availbale
  • 26. June 2018 - Featured blog on Medium
    "Hierarchical Context in Conversations for Spoken Language Understanding"
  • 3. June 2018 - One of the related paper is accepted at Interspeech 2018!
    "Conversational Analysis using Utterance-level Attention-based Bidirectional Recurrent Neural Networks"
  • 25. May 2018 - Initial release of the Discourse-Wizard live web-demo

Neural Models



The context-based approach, which can take the preceding utterance (or utterances) into account, is crucial for language understanding modules in any dialogue engines. We use the hierarchical recurrent neural networks (RNN) to model such a context for conversational analysis as shown in the followind slides.

Hierarchical Context-based Dialogue Act Recognition using RNNs

Code

Code will be available soon, stay tuned...

References

Bothe, C., Magg, S., Weber, C., and Wermter, S. (2018).
Discourse-Wizard: Discovering Deep Discourse Structure in your Conversation with RNNs .
arXiv:1806.11420 [cs.CL]


Bothe, C., Magg, S., Weber, C., and Wermter, S. (2018).
Conversational Analysis using Utterance-level Attention-based Bidirectional Recurrent Neural Networks.
Proceedings of INTERSPEECH 2018.


Bothe, C., Weber, C., Magg, S., and Wermter, S. (2018).
A Context-based Approach for Dialogue Act Recognition using Simple Recurrent Neural Networks.
Proceedings of the Language Resources and Evaluation Conference (LREC-2018).

Footprint

not supported in your browser Updated on 30.10.2018. Based on the user/visitor requests sent to the server. Generated by Google Map using gmplot.