Detecting Polarized Politicians in Online Discussions with Deep Learning

Description

We aim to identify political individuals (e.g., the US presidents) who are discussed in online conversations (e.g., Reddit) and receive polarized opinions. Detecting these polarized politicians is crucial for understanding which ones may be prone to discussions involving hate speech, stereotypes, and other harmful content. This information helps us recognize potential areas where such negative behaviors might occur. Our project aims to address this issue by developing computational tools utilizing deep learning.

We seek to implement and extend the findings of the study [1] that is based on Graph Neural Networks. The authors discovered the polarized concepts (e.g., abortion and gun control) in Reddit Politosphere [2] across different subreddits.

Our focus is on detecting polarized political individuals on Reddit Politosphere. Our research will explore the identification of polarized politicians over time and across different subreddits, contributing to a better understanding of online polarization toward politics and offering insights to mitigate their adverse effects.

Keywords

Polarized Politicians, Social media, Deep Learning, Graph Neural Networks

Proposed Steps

1- Identifying the names of politicians in Reddit Politosphere using NER tools

2- Employing Graph Neural Networks (as described in the paper [1]) for identifying the polarized politicians on Reddit Politosphere

3- Conducting analysis over time and over subreddits to understand the public opinion about the politicians (e.g., the US presidents)

References

[1] V. Hofmann, X. Dong, J. Pierrehumbert, and H. Schuetze, "Modeling Ideological Salience and Framing in Polarized Online Groups with Graph Neural Networks and Structured Sparsity," in Findings of the Association for Computational Linguistics: NAACL 2022, Seattle, United States, 2022, pp. 536–550. Association for Computational Linguistics.

[2] V. Hofmann, H. Schütze, and J. B. Pierrehumbert, “The Reddit Politosphere: A Large-Scale Text and Network Resource of Online Political Discourse”, ICWSM, vol. 16, no. 1, pp. 1259-1267, May 2022.