Dialogue Act-based Breakdown Detection in Negotiation Dialogues

This repository contains the proposed dataset and the negotiation interface used to collect our data. It also contains the official implementation.

Dataset

Please unzip data.zip. The unzipped data: data.json is our proposed JI dataset.

Negotiation Interface

Implementations regarding our negotiation interface are available in the interface folder.

How to extract samples

We provide a helper script: helper/negotiation_ji.py that easily enables users to extract the attributes of each dialogue in the JI dataset.

Any utterances in a dialogue can be extracted with the following procedures:

1. Load dialogues using read_ji_negotiations(filename=/path/to/data.json/). This will return the list of the Negotiation object.

2. Get the list of Comment object from each Negotiation object.

3. Comment.body has an utterance from a certain user.

Citation

@inproceedings{yamaguchi-etal-2021-dialogue,
    title = "Dialogue Act-based Breakdown Detection in Negotiation Dialogues",
    author = "Yamaguchi, Atsuki  and
      Iwasa, Kosui  and
      Fujita, Katsuhide",
    booktitle = "Proceedings of the 16th Conference of the European Chapter of the Association for Computational Linguistics: Main Volume",
    month = apr,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.eacl-main.63",
    pages = "745--757",
    abstract = "Thanks to the success of goal-oriented negotiation dialogue systems, studies of negotiation dialogue have gained momentum in terms of both human-human negotiation support and dialogue systems. However, the field suffers from a paucity of available negotiation corpora, which hinders further development and makes it difficult to test new methodologies in novel negotiation settings. Here, we share a human-human negotiation dialogue dataset in a job interview scenario that features increased complexities in terms of the number of possible solutions and a utility function. We test the proposed corpus using a breakdown detection task for human-human negotiation support. We also introduce a dialogue act-based breakdown detection method, focusing on dialogue flow that is applicable to various corpora. Our results show that our proposed method features comparable detection performance to text-based approaches in existing corpora and better results in the proposed dataset.",
}

License

MIT License