Multilingual AMR Discourse Coherence: Representation of a Social Movement on Social Media

ACL ARR 2025 February Submission4881 Authors

Abstract: Combining AMR with social sciences and discourse coherence is a new interdisciplinary research focus that has potential to explore the meaning of online public discourse through parsed sentence structure of the Abstract Meaning Representation (AMR) graph. The objective of this research is to provide a better sense of how the anti-femicide movement in Mexico is circulated on social media by understanding the narrative structure of post comments. This work attempts to reproduce a similar research framework as \citeposs{pournaki2024graph} and their political discourse narratives on social media text comments about the anti-femicide movement in Mexico. It uses a three step process to translate the data into all English, parse the data into an AMR graph, and analyze that graph to understand the narrative surrounding the movement. It found that the process is almost generalizable, but only provides legitimate findings with specific data types.

Paper Type: Short

Research Area: Computational Social Science and Cultural Analytics

Research Area Keywords: Discourse and Pragmatics, Machine Translation, Multilingualism and Cross-Lingual NLP, Computational Social Science and Cultural Analytics

Contribution Types: Reproduction study, Publicly available software and/or pre-trained models, Data analysis

Languages Studied: Spanish, English

Submission Number: 4881