Conflict & Sentiment Analysis

BigQuery · Python · scikit-learn · Power BI

Incremental pipeline classifying global conflict and scoring sentiment on the GDELT Event 2.0 dataset, refreshed every 15 minutes into a Power BI dashboard.

Use case

Replicates the monitoring feed a geopolitical or newsroom analyst would use: a near-real-time view of international events, what kind of conflict they represent, and how the tone of coverage is shifting by country.

Architecture

The pipeline pulls new GDELT Event 2.0 records from BigQuery on a 15-minute cycle, applies feature engineering, and classifies events as verbal or material and internal or external conflicts using supervised and unsupervised scikit-learn models. Sentiment is scored from the Goldstein scale and AvgTone, and the results land in a Power BI dashboard showing sentiment, event magnitude and country involvement.