This is a PyTorch implementation of the paper: Domain-Aware Dialogue State Tracker for Multi-Domain Dialogue Systems. by Vevake Balaraman and Bernardo Magnini.

In task-oriented dialogue systems the dialogue state tracker (DST) component is responsible for predicting the state of the dialogue based on the dialogue history. Current DST approaches rely on a predefined domain ontology, a fact that limits their effective usage for large scale conversational agents, where the DST constantly needs to be interfaced with ever-increasing services and APIs. Focused towards overcoming this drawback, we propose a domain-aware dialogue state tracker, that is completely data-driven and it is modeled to predict for dynamic service schemas including zero-shot domains. Unlike approaches that propose separate models for prediction of intents, requested slots, slot status, categorical slots and non-categorical slots, we propose a single model in an end-to-end architecture. The proposed model also utilizes domain and slot information to extract both domain and slot specific representations from a given dialogue, and then uses such representations to predict the values of the corresponding slot in a given domain. Integrating this mechanism with a pretrained language models, our approach can effectively learn semantic relations and effectively perform zero-shot tracking for domains not present in training.

The bibtex is below.

@article{balaraman2020domainaware,
      title={Domain-Aware Dialogue State Tracker for Multi-Domain Dialogue Systems},
      author={Vevake Balaraman and Bernardo Magnini},
      year={2020},
      eprint={2001.07526},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

The Schema Guided Dataset (SGD) is used for the experiments.

1. Configure the directories and BERT model locations in config.py

2. Create a schema dictionary for the dataset.

   python create_schema_dict.py

3. create schema encodings

   python encode_schema.py

4. Train the model

   python train.py

5. Test the final model

   python test.py

Please feel free to contact me at [email protected] for any queries.