Detecting and Explaining Conflicting Narratives: Computational Reconciliation of Divergent Political Text Streams

While previous work (Hu & Kearney, 2020) has used computational methods to surface group differences in discourse, and Gavras et al. (2022) have tracked policy convergence or divergence, there’s little on how and why conflicting frames or narratives co-exist or shift in real time. This idea leverages transformer-based multi-view learning to detect not just what is being discussed, but where direct contradictions or framing battles occur, and uses causal text analysis to infer the sources or triggers of these conflicts (e.g., major events, polarization, coordinated campaigns). It could draw on techniques from fake news detection (Sowmya H.K. & Anandhi R.J., 2024) but applies them to political narrative conflict, not just veracity. This would enable researchers to untangle the roots and evolution of political polarization or misinformation, offering a new lens for studying political communication dynamics.

References:

1. Gendered Tweets: Computational Text Analysis of Gender Differences in Political Discussion on Twitter. Lingshu Hu, M. Kearney (2020). Journal of Language and Social Psychology.
2. Convergence of European security and defense preferences? A quantitative text analysis of strategy papers, 1994–2018. Konstantin Gavras, M. Mader, Harald Schoen (2022). European Union Politics.
3. Comparative Analysis of Deep Learning Techniques for Fake News Detection. Sowmya H.K., Anandhi R.J. (2024). 2024 3rd Edition of IEEE Delhi Section Flagship Conference (DELCON).