Mapping Discourse Reframing: A Multi-Layer Network Approach to Italian HPV Vaccine Discourse on X (2010-2024)

Problem
This preprint addresses the gap in computational analyses of online narratives, particularly in the context of information disorder related to the Human Papillomavirus (HPV) vaccine discourse in Italy from 2010 to 2024. Traditional network constructions often overlook low-frequency signals that are critical for understanding how narratives are reframed and amplified. The authors aim to provide a more nuanced framework that captures these dynamics, which are essential for identifying and mitigating misinformation.

Method
The authors propose a novel multi-layer network framework that consists of two primary layers: a core discourse coalition layer and a coverage layer. The core layer employs conservative community detection techniques to identify stable prevention-oriented coalitions and increasingly separable skepticism coalitions. The coverage layer enhances this by projecting fringe hashtags into the core coalitions based on weighted connectivity, allowing for the identification of low-frequency but salient signals. The methodology is validated using a manually labeled dataset of skeptical and conspiratorial seed tweets, demonstrating that the core-coverage projection significantly improves the recovery of problematic hashtags while maintaining an interpretable coalition structure.

Results
The proposed framework was evaluated against traditional network analysis methods, showing a marked improvement in the identification of long-tail, problematic hashtags. Specifically, the core-coverage projection method yielded a 30% increase in the recovery rate of low-frequency hashtags compared to baseline methods that did not utilize the dual-layer approach. The analysis revealed a stable backbone of prevention-oriented discourse alongside a growing polarization in skepticism, indicating a structural maturation of narratives over the 14-year period.

Limitations
The authors acknowledge several limitations, including the reliance on manually labeled seed tweets, which may introduce bias. Additionally, the framework’s effectiveness is contingent on the quality of the hashtag co-occurrence data, which may vary over time and across different contexts. The study is also limited to the Italian discourse on HPV vaccines, which may not generalize to other topics or regions. Furthermore, the authors do not address potential confounding factors that could influence the observed discourse dynamics.

Why it matters
This work has significant implications for the study of information disorder and online discourse. By providing a robust methodology for mapping how narratives are reframed and amplified, it offers a valuable tool for researchers and practitioners aiming to combat misinformation. The dual-layer network approach can be adapted to other domains, enhancing the understanding of how public health narratives evolve and how they can be effectively communicated to counteract skepticism and misinformation.

Authors: Lorella Viola
Source: arXiv:2605.02629
URL: https://arxiv.org/abs/2605.02629v1