This is the code for the paper Exploiting Personal Characteristics of Debaters for Predicting Persuasiveness.
Khalid Al-Khatib, Michael Völske, Shahbaz Syed, Nikolay Kolyada, and Benno Stein
@InProceedings{stein:2020m,
author = {Khalid Al-Khatib and Michael V\"olske and Shahbaz Syed and Nikolay Kolyada and Benno Stein},
booktitle = {58th Annual Meeting of the Association for Computational Linguistics (ACL 2020)},
month = jul,
pages = {7067-7072},
publisher = {Association for Computational Linguistics},
site = {Seattle, USA},
title = {{Exploiting Personal Characteristics of Debaters for Predicting Persuasiveness}},
url = {https://www.aclweb.org/anthology/2020.acl-main.632},
year = 2020
}
The provided code contains the set of PySpark jobs used for preprocessing and construction of final datasets from Reddit Crawl corpus. The jobs under pyspark/ folder have to be packed into a zip archive and run with commands listed in run.sh. Respective build commands in the Makefile can be used.
The provided Jupyter notebooks demostrate the main steps taken for two tasks studied in the paper: load the preprocessed datasets, prepare feature dictionaries, train models and do feature selection. (Task 1: notebooks 1-3, Task 2: notebook 4).