Latent Geometry of Taste: Scalable Low-Rank Matrix Factorization for Recommender Systems
Joshua Salako · 2026
Read Latent Geometry of Taste: Scalable Low-Rank Matrix Factorization for Recommender Systems by Joshua Salako (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-03466.txt.
Description
Joshua Salako, working alone, studies how recommender systems learn the hidden shape of user taste. He uses the MovieLens 32M dataset, a large public record of movie ratings, and a fast parallel version of alternating least squares, a standard fitting method. The goal is to handle sparse data, since most users rate only a tiny share of all films.
Careful tuning showed that smaller, constrained models generalized better than larger ones, balancing prediction error and ranking quality. Maps of the learned space revealed genre clusters that emerged on their own from ratings alone. Salako also tested cold starts, where a user is new, and added a dial to trade popularity against personal fit. Code for the work is public.
Catalog record
| CREATOR | Joshua Salako |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 23 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-03466