Polarization Index Calculator
A computational tool for computing polarization indices from structured discourse data, enabling quantitative comparison of polarization levels across corpora, time periods, and institutional contexts.
Research Question
Can we quantify the degree of polarization in a given corpus of public statements?
Data Source
Parliamentary records, media transcripts
Limitations
Requires structured input, does not capture visual rhetoric
Demo not available
Note: This tool is currently in development and not yet available for use. The documentation below describes the planned functionality. Expected availability: Q3 2026.
Overview
The Polarization Index Calculator (PIC) is a research prototype that computes composite polarization scores from structured discourse data. It operationalizes the three-dimensional polarization model described in DuMonde's Polarization Diagnosis Method, producing quantitative indices for affective, cognitive, and structural polarization that can be tracked over time and compared across contexts.
The tool takes structured input — tagged discourse units with speaker metadata, temporal markers, and topical classifications — and produces a set of normalized indices on a 0-to-1 scale. These indices are not intended as absolute measures but as relative indicators that enable comparison: Is polarization in this parliamentary session higher than last year? Does media discourse show more affective or cognitive polarization?
Technical Approach
The calculator processes discourse data stored in a SQLite database, applying a battery of computational measures for each polarization dimension. Affective polarization is estimated through sentiment divergence between identified speaker clusters. Cognitive polarization is measured by semantic distance between group-level topic models. Structural polarization is computed using network modularity scores derived from reply and citation patterns.
Results are visualized through interactive Plotly dashboards that display index trends, dimensional breakdowns, and comparative views. Users can filter by time period, topic, speaker group, or institutional context.
Current Status and Limitations
The PIC is in alpha stage. It currently requires pre-processed, structured input conforming to a specific schema, which limits accessibility for users without data preparation capabilities. The tool does not analyze visual rhetoric — images, video, or graphic design elements — which are increasingly significant vectors of polarization in digital discourse. Future development will focus on expanding input compatibility, adding visual analysis modules, and validating index reliability across diverse linguistic and cultural contexts.