کنفرانس بین المللی محاسبات نرم

صفحه اصلی / ششمین کنفرانس بین المللی محاسبات نرم

How Spotify’s Recommendation Algorithms Power Personalized Playlists: A Comparative Analysis with YouTube Music and Pandora.

نویسندگان :

Nima Nikrouz¹ Ali Shahnazi² Abdorreza Hesam Mohseni³

1- 'دانشگاه گیلان - دانشکده فنی و مهندسی شرق گیلان 2- دانشگاه گیلان - دانشکده فنی و مهندسی شرق گیلان 3- 'دانشگاه گیلان - دانشکده فنی و مهندسی شرق گیلان

کلمات کلیدی :

Recommendation Algorithms،Personalization،Spotify،Collaborative Filtering،Content-Based Filtering،NLP

چکیده :

Spotify’s recommendation system, a hybrid model combining collaborative filtering, content-based analysis, and Natural Language Processing (NLP), drives 70% of user listening and powers 4 billion personalized playlists monthly, shaping music discovery for 626 million active users worldwide. This article analyzes its mechanics, revealing strengths like 85% accuracy in Discover Weekly (reaching 2.3 billion streams) and 25-minute average daily sessions, while exposing weaknesses such as 60% playlist repetition and 40% echo chamber risk. Compared to YouTube Music’s 80% video-driven recommendations and Pandora’s 70% genre-tagging precision, Spotify leads in engagement but lags in diversity. Real-world data shows 1 in 3 users skip repetitive tracks, and 55% crave more serendipity. Proposed improvements—mood-aware filtering (boosting variety by 30% in trials) and adjustable shuffle sliders (reducing repetition by 45%)—address these gaps. Drawing on 2024–2025 industry metrics, this study affirms Spotify’s dominance in personalized streaming while offering data-backed enhancements to reduce bias, increase discovery, and elevate user satisfaction across 100 million tracks.