Document type
Bachelor thesisPublication date
Publication license
Please use this identifier to cite or link to this item: https://hdl.handle.net/2445/130481
Word2vec embeddings for playlist recommendation
Journal Title
Authors
Director/Tutor
Journal ISSN
Volume Title
Related resource
Abstract
[en] We present an ML approach to musical playlist recommendation. Using the algorithm Word2Vec, a shallow two-layer neural network trained to reconstruct linguistic context of words, we have created several embeddings using tracks and playlist titles as words of an artificial vocabulary. Some experiments with different trade-offs between the diversity and the popularity of songs in playlists are analyzed and discussed. By means of combining a tracks embedding and a titles embedding our recommender has reached 19 percent of accuracy. Our model has been created and trained using the MPD (million playlists dataset) given by Spotify as part of the RecSys Challenge 2018.
Description
Treballs Finals de Grau d'Enginyeria Informàtica, Facultat de Matemàtiques, Universitat de Barcelona, Any: 2018, Director: Santi Seguí Mesquida i Guillem Pascual i Guinovart
Subject (English)
Citation
Citation
BACH VALLS, Anna. Word2vec embeddings for playlist recommendation. [consulted: 14 of August of 2026]. Available at: https://hdl.handle.net/2445/130481