Of Spiky SVDs and Music Recommendation
This addresses incremental improvements in understanding embedding stability for music recommendation systems.
The paper investigates spiking formations in embedding spaces from truncated SVD in music recommendation, proving their origin in item popularity communities, and applies this to predict changes in top-k similar items over time with new data.
The truncated singular value decomposition is a widely used methodology in music recommendation for direct similar-item retrieval or embedding musical items for downstream tasks. This paper investigates a curious effect that we show naturally occurring on many recommendation datasets: spiking formations in the embedding space. We first propose a metric to quantify this spiking organization's strength, then mathematically prove its origin tied to underlying communities of items of varying internal popularity. With this new-found theoretical understanding, we finally open the topic with an industrial use case of estimating how music embeddings' top-k similar items will change over time under the addition of data.