clef2019-prosody

This repository contains the dataset release to accompany the CLEF 2019 paper titled

"Using Audio Transformations to Improve Comprehension in Voice Question Answering" by Aleksandr Chuklin, Aliaksei Severyn, Johanne R. Trippas, Enrique Alfonseca, Hanna Silen, and Damiano Spina.

Please, use the following citation

@inproceedings{clef2019prosody,
  title={{Using Audio Transformations to Improve Comprehension in Voice Question Answering}},
  author = {Aleksandr Chuklin and
            Aliaksei Severyn and
            Johanne R. Trippas and
            Enrique Alfonseca and
            Hanna Silen and
            Damiano Spina},
  booktitle={{Conference and Labs of the Evaluation Forum (CLEF)}},
  year={2019},
  location = {Lugano, Switzerland}
}

You may also refer to the extended version on ArXiv: https://arxiv.org/abs/1806.03957

Input Data

The data used for ratings comes from the Stanford Question Answering Dataset (SQuAD) (distributed under the CC BY-SA 4.0 license).

Rating Interface for Crowd Workers

Experiments were performed under Ethics Application BSEH 10-14 at RMIT University.

Example Audios

See media/ folder for some example. You may also download more files by following the links in data/crowdsourcing_ratings.csv.