@inproceedings{thompson-etal-2025-discourse,
    title = "Does discourse structure help action prediction? A look at Correction Triangles.",
    author = "Thompson, Kate  and
      Chaturvedi, Akshay  and
      Asher, Nicholas",
    editor = "Evang, Kilian  and
      Kallmeyer, Laura  and
      Pogodalla, Sylvain",
    booktitle = "Proceedings of the 16th International Conference on Computational Semantics",
    month = sep,
    year = "2025",
    address = {D{\"u}sseldorf, Germany},
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/author-page-yu-wang-polytechnic/2025.iwcs-main.16/",
    pages = "166--174",
    ISBN = "979-8-89176-316-6",
    abstract = "An understanding of natural language corrections is essential for artificial agents that are meant to collaborate and converse with humans. We present some preliminary experiments using language-to-action models investigating whether discourse structure, in particular Correction relations, improves the action prediction capabilities of language-to-action models for simple block world tasks. We focus on scenarios in which a model must correct a previous action, and present a corpus of synthetic dialogues to help explain model performance."
}