Analyzing Persuasion Strategies of Debaters on Social Media

Analyzing Persuasion Strategies of Debaters on Social Media - Dataset

This dataset contains 3,801 debaters from Reddit, their comment, and their persuasion effectiveness. The debates originate from the subreddit Chage my View and are extracted from the Webis CMV dataset (Al Khatib et al., 2020). The dataset consists of three files in JSON Lines (.jsonl) format.

Content of the Dataset
+-- Reddit Debaters
| +-- README.md # This information + file format description
| +-- debaters.jsonl # Minimal dataset with only the (source) comment text and persuasiveness
| +-- debaters-full.jsonl # All debater-level datapoints, computed an retrieved from Reddit
| +-- comments.jsonl # All comments and comment-level datapoints for each debater

Cite
@inproceedings{wiegmann:2022,
title = "Re-framing Incremental Deep Language Models for Dialogue Processing with Multi-task Learning",
author = "Wiegmann, Matti and Al-Khatib, Khalid and Khanna, Vishal and Stein, Benno",
booktitle = "Proceedings of the 29th International Conference on Computational Linguistics",
month = oct,
year = "2022",
address = "Gyeongju, Republic of Korea",
publisher = "International Committee on Computational Linguistics",
}