A Multi-Level Benchmark for Causal Language Understanding in Social Media Discourse

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ding, Xiaohan, Ping, Kaike, Çarık, Buse, Rho, Eugenia
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914049238761472
author Ding, Xiaohan
Ping, Kaike
Çarık, Buse
Rho, Eugenia
author_facet Ding, Xiaohan
Ping, Kaike
Çarık, Buse
Rho, Eugenia
contents Understanding causal language in informal discourse is a core yet underexplored challenge in NLP. Existing datasets largely focus on explicit causality in structured text, providing limited support for detecting implicit causal expressions, particularly those found in informal, user-generated social media posts. We introduce CausalTalk, a multi-level dataset of five years of Reddit posts (2020-2024) discussing public health related to the COVID-19 pandemic, among which 10120 posts are annotated across four causal tasks: (1) binary causal classification, (2) explicit vs. implicit causality, (3) cause-effect span extraction, and (4) causal gist generation. Annotations comprise both gold-standard labels created by domain experts and silver-standard labels generated by GPT-4o and verified by human annotators. CausalTalk bridges fine-grained causal detection and gist-based reasoning over informal text. It enables benchmarking across both discriminative and generative models, and provides a rich resource for studying causal reasoning in social media contexts.
format Preprint
id arxiv_https___arxiv_org_abs_2509_16722
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Multi-Level Benchmark for Causal Language Understanding in Social Media Discourse
Ding, Xiaohan
Ping, Kaike
Çarık, Buse
Rho, Eugenia
Computation and Language
Understanding causal language in informal discourse is a core yet underexplored challenge in NLP. Existing datasets largely focus on explicit causality in structured text, providing limited support for detecting implicit causal expressions, particularly those found in informal, user-generated social media posts. We introduce CausalTalk, a multi-level dataset of five years of Reddit posts (2020-2024) discussing public health related to the COVID-19 pandemic, among which 10120 posts are annotated across four causal tasks: (1) binary causal classification, (2) explicit vs. implicit causality, (3) cause-effect span extraction, and (4) causal gist generation. Annotations comprise both gold-standard labels created by domain experts and silver-standard labels generated by GPT-4o and verified by human annotators. CausalTalk bridges fine-grained causal detection and gist-based reasoning over informal text. It enables benchmarking across both discriminative and generative models, and provides a rich resource for studying causal reasoning in social media contexts.
title A Multi-Level Benchmark for Causal Language Understanding in Social Media Discourse
topic Computation and Language
url https://arxiv.org/abs/2509.16722