STEntConv: Predicting Disagreement with Stance Detection and a Signed Graph Convolutional Network

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lorge, Isabelle, Zhang, Li, Dong, Xiaowen, Pierrehumbert, Janet B.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929289277997056
author Lorge, Isabelle
Zhang, Li
Dong, Xiaowen
Pierrehumbert, Janet B.
author_facet Lorge, Isabelle
Zhang, Li
Dong, Xiaowen
Pierrehumbert, Janet B.
contents The rise of social media platforms has led to an increase in polarised online discussions, especially on political and socio-cultural topics such as elections and climate change. We propose a simple and novel unsupervised method to predict whether the authors of two posts agree or disagree, leveraging user stances about named entities obtained from their posts. We present STEntConv, a model which builds a graph of users and named entities weighted by stance and trains a Signed Graph Convolutional Network (SGCN) to detect disagreement between comment and reply posts. We run experiments and ablation studies and show that including this information improves disagreement detection performance on a dataset of Reddit posts for a range of controversial subreddit topics, without the need for platform-specific features or user history.
format Preprint
id arxiv_https___arxiv_org_abs_2403_15885
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle STEntConv: Predicting Disagreement with Stance Detection and a Signed Graph Convolutional Network
Lorge, Isabelle
Zhang, Li
Dong, Xiaowen
Pierrehumbert, Janet B.
Computation and Language
The rise of social media platforms has led to an increase in polarised online discussions, especially on political and socio-cultural topics such as elections and climate change. We propose a simple and novel unsupervised method to predict whether the authors of two posts agree or disagree, leveraging user stances about named entities obtained from their posts. We present STEntConv, a model which builds a graph of users and named entities weighted by stance and trains a Signed Graph Convolutional Network (SGCN) to detect disagreement between comment and reply posts. We run experiments and ablation studies and show that including this information improves disagreement detection performance on a dataset of Reddit posts for a range of controversial subreddit topics, without the need for platform-specific features or user history.
title STEntConv: Predicting Disagreement with Stance Detection and a Signed Graph Convolutional Network
topic Computation and Language
url https://arxiv.org/abs/2403.15885