{"ID":2891082,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2507.18800","arxiv_id":"2507.18800","title":"Semantic IDs for Music Recommendation","abstract":"Training recommender systems for next-item recommendation often requires unique embeddings to be learned for each item, which may take up most of the trainable parameters for a model. Shared embeddings, such as using content information, can reduce the number of distinct embeddings to be stored in memory. This allows for a more lightweight model; correspondingly, model complexity can be increased due to having fewer embeddings to store in memory. We show the benefit of using shared content-based features ('semantic IDs') in improving recommendation accuracy and diversity, while reducing model size, for two music recommendation datasets, including an online A/B test on a music streaming service.","short_abstract":"Training recommender systems for next-item recommendation often requires unique embeddings to be learned for each item, which may take up most of the trainable parameters for a model. Shared embeddings, such as using content information, can reduce the number of distinct embeddings to be stored in memory. This allows f...","url_abs":"https://arxiv.org/abs/2507.18800","url_pdf":"https://arxiv.org/pdf/2507.18800v1","authors":"[\"M. Jeffrey Mei\",\"Florian Henkel\",\"Samuel E. Sandberg\",\"Oliver Bembom\",\"Andreas F. Ehmann\"]","published":"2025-07-24T20:48:02Z","proceeding":"cs.IR","tasks":"[\"cs.IR\",\"cs.LG\"]","methods":"[]","has_code":false}