Lo scopo principale è stimare il sentiment e i segnali di brand perception combinando più fonti: trascrizioni dei video, commenti social, titoli/descrizioni editoriali e frame/thumbnail. La tesi realizza una pipeline multimodale (testo+audio+immagine) e correla gli indici ottenuti con metriche di performance (CTR, view-through, completion rate) per costruire un immediato “brand-lift proxy”.

Argomento principale: Data Science, NLP/NLU, Computer Vision, Audio sentiment, Metriche editoriali.

Corso di studio e requisiti candidati: Informatica, Ingegneria Informatica. Solida base in Python e librerie ML; gradite nozioni di analisi social.

Sede tirocinio: Napoli e Milano.

_

The main goal is to estimate sentiment and brand perception signals by combining multiple sources: video transcripts, social comments,



editorial titles/descriptions, and frames/thumbnails.

The thesis will implement a multimodal pipeline (text + audio + image) and correlate the resulting indices with performance metrics (CTR, view-through, completion rate) to build a simple “brand-lift proxy”.

Main Topic: Data Science, NLP/NLU, Computer Vision, Audio sentiment, Editorial metrics.

Course of Study and Candidate Requirements: Computer Science or Computer Engineering. A solid foundation in Python and ML libraries; knowledge of social analysis is a plus.

Internship Location: Naples and Milan.

📌 Multimodal Sentiment & Brand Perception For Media Content Napoli
🏢 Kineton
📍 Napoli