Skip to content

DeliData

A dataset for deliberation in multi-party problem solving

Small groups participate in a group discussion to solve the Wason card selection task.


The dataset was collected both in a controlled in-lab experiment and in-the-wild data collection on MechanicalTurk


50 dialogues annotated with a novel annotation schema that captures deliberation cues

Annotation Scheme
Label Distribution

How to access DeliData

2. Use delitoolkit. Documentation.

pip install delitoolkit
from delitoolkit.delidata import DeliData
deli_corpus = DeliData().corpus

3. Use HuggingFace datasets. Dataset Card

from datasets import load_dataset
dataset = load_dataset("gkaradzhov/DeliData")

Citation

@article{karadzhov2023delidata,
    title={DeliData: A dataset for deliberation in multi-party problem solving},
    author={Karadzhov, Georgi and Stafford, Tom and Vlachos, Andreas},
    journal={Proceedings of the ACM on Human-Computer Interaction},
    volume={7},
    number={CSCW2},
    pages={1--25},
    year={2023},
    publisher={ACM New York, NY, USA}
  }

This dataset was collected with the support of Cambridge Language Sciences, Cambridge University Press, and the Isaac Newton Trust.