🎬 From Screen to Heart – How Do Emotions Shape the Cinematic Landscape?

📍 EPFL – Master in Data Science, Year 1 (2024)
👥 Team: Sofia Taouhid, Lina Sadgal, Yasmine Tligui, Guillaume Vitalis
🏆 Award: Selected among the Top 10 Projects (out of 100+ teams)
🔗 Project Website: Demo
🔗 Code Repository: GitHub

This project investigates the emotional dimensions of storytelling using a large corpus of over 11,000 movie summaries enriched with 450,000 IMDb reviews.

We used HuggingFace’s Emotion DistilRoBERTa model to extract primary emotions — joy, sadness, fear, anger, surprise, disgust, neutrality — then applied statistical correlations, PCA, and K-Means clustering to map emotional trends across genres and narratives.

Inspired by Inside Out, our project features a visual storytelling approach, integrating emotion-based visualizations like word clouds and heatmaps to offer an engaging exploration of genre-emotion relationships.

🛠 Tools & Libraries:

🧠 Techniques: