Harnessing High-Level Song Descriptors towards Natural Language-Based Music Recommendation

TL;DR

This paper evaluates how effectively language models utilize high-level song descriptors for natural language-based music recommendation, showing improved performance with domain-specific fine-tuning.

Contribution

It introduces a formulation of music recommendation as a dense retrieval task and assesses language models' effectiveness in this context.

Findings

Performance improves with fine-tuning for language similarity.

Better mapping of descriptions to high-level song descriptors.

Enhanced recommendation accuracy with domain-specific data.

Abstract

Recommender systems relying on Language Models (LMs) have gained popularity in assisting users to navigate large catalogs. LMs often exploit item high-level descriptors, i.e. categories or consumption contexts, from training data or user preferences. This has been proven effective in domains like movies or products. However, in the music domain, understanding how effectively LMs utilize song descriptors for natural language-based music recommendation is relatively limited. In this paper, we assess LMs effectiveness in recommending songs based on user natural language descriptions and items with descriptors like genres, moods, and listening contexts. We formulate the recommendation task as a dense retrieval problem and assess LMs as they become increasingly familiar with data pertinent to the task and domain. Our findings reveal improved performance as LMs are fine-tuned for general language similarity, information retrieval, and mapping longer descriptions to shorter, high-level descriptors in music.