ThemeScope

I'm glad to share that our paper "ThemeScope: A quantitative thematic analysis for depicting social representations in digital arenas" has been published in Journal of Information Science (SAGE), co-authored with Michelangelo Misuraca and Maria Spano.

ThemeScope is a quantitative approach designed to detect and visualise social representations, the shared systems of meaning that circulate in large-scale digital text, directly from the words people use online. Rather than relying on manual coding, it turns a corpus of documents into an interpretable map of themes.

How it works

The method reconstructs meaning from language through a transparent pipeline: it annotates the text, builds a sentence-level word co-occurrence network, and identifies thematic communities within it. Each theme is then positioned on a strategic diagram using two indicators grounded in Social Representation Theory:

• Prototypical Salience Index (PSI): captures how structurally central and frequent a theme is (its anchoring in the collective representation).
• Concreteness Score (CS): measures how concrete and imageable a theme's vocabulary is (its objectification).

The themescopeR package

The approach is fully implemented in the open-source R package themescopeR, which covers the whole workflow, from importing documents (CSV, XLSX, TXT, ZIP) and linguistic annotation, to community detection (walktrap, Louvain, Leiden), computation of the PSI and CS indicators, and visualisation of strategic diagrams and networks, either from the console or through an optional Shiny interface.