Replicating and extending cybersecurity risk analysis: A methodological note

Abstract

This methodological note details the technical infrastructure developed to replicate and extend the study by Florackis et al. (2023) on cybersecurity risk disclosures. Adopting an opensource approach, the project provides a transparent view into the mechanics of textual analysis. I present two analytical pipelines: a replication using font-aware extraction to preserve document structure, and an extension using Latent Dirichlet Allocation (LDA) to uncover latent themes. The manuscript details critical steps in data acquisition, cleaning, dictionary creation, and the shift from keyword counting to unsupervised topic modeling