Non-negative matrix factorization algorithms generally improve topic model fits
Peter Carbonetto, Abhishek Sarkar, Zihao Wang, Matthew Stephens · 2026
Read Non-negative matrix factorization algorithms generally improve topic model fits by Peter Carbonetto, Abhishek Sarkar, Zihao Wang, Matthew Stephens (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/pmc-13144203.txt.
Description
Peter Carbonetto, Abhishek Sarkar, Zihao Wang, and Matthew Stephens published a 2026 study of a fast way to fit topic models to large text collections. A topic model sorts texts by the themes they mix. The best-fit search for such a model has a twin in a known matrix-splitting task, non-negative matrix factorization. That twin link was known informally, but no one had used it to fit topic models.
Recent advances in the matrix-splitting tools can be turned to the topic task, the team shows. The fits arrive fast, and they land better than the standard topic tools provide. The team also proves the link in strict formal form. One standard topic tool turns out to be the same, at its core, as the classic matrix-splitting rule. Their tools ship in a free code package for wide use.
Catalog record
| CREATOR | Peter Carbonetto, Abhishek Sarkar, Zihao Wang, Matthew Stephens |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 130 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |