An automatic framing approach to analyse newspaper and press language

This study introduces an automated framework for framing analysis to interpret media discourse on public policy. The approach identifies structured narratives that can shape public perception by detecting frames as clusters of co-occurring terms. The method applies the Leiden algorithm, enhancing frame detection by grouping related terms in a network, where nodes represent terms and edges their co-occurrences. The texts are modelled as a weighted graph, and the algorithm's modularity-based clustering highlights cohesive frames. Applied to Italian media coverage of ``Reddito di Cittadinanza'' (RdC) from 2018 to 2023, the analysis revealed 21 frames covering economic, civic, and political aspects, aligning with legislative milestones and reflecting shifts in the public interest. This case study demonstrates the framework's effectiveness for large-scale media analysis and its broader applicability in policy discourse research.