# Distributional word classes (PPMI-SVD embeddings, k-means)

> Run 5 times on the site, with every result.

Source: https://mysteriesdept.com/methods/MTH08JFXDCR6J0N/distributional-word-classes-ppmi-svd-embeddings-k-means  
Publisher: Mysteries Department  
Licence: CC BY 4.0 (credit Mysteries Department and link the page)  

---

[Methods](https://mysteriesdept.com/methods)[NLP / Copiale-style context clustering](https://mysteriesdept.com/methods?kind=language-and-text)Distributional word classes (PPMI-SVD embeddings, k-means)

Method · NLP / Copiale-style context clustering

# Distributional word classes (PPMI-SVD embeddings, k-means)

**Nlp / copiale-style context clustering** Version **1.0** Licence **MIT5** runsUsed on **1** mystery

## How it works

Compared with: Latin of the manuscript's genres, slot-grammar text and the manuscript shuffled across pages, all in the same lines.

## Code Python

[Download module (.py)](https://mysteriesdept.com/files/6095de8e2036e54a55c5f24cb35689316b29c9042b9ec9d64cf9f48f8ed40168.py)`methods_ind.py`[View source](https://mysteriesdept.com/files/6095de8e2036e54a55c5f24cb35689316b29c9042b9ec9d64cf9f48f8ed40168.py)

## Runs and results 5 runs

| Run | Mystery | Measured | Result | Against | Date |
|---|---|---|---|---|---|
| RUN-095 | Voynich | Nmi class line position | 0.0164 | Latin 0.0029 · slot 0.0034 · shuffled 0.0007 | 22 Oct |
| RUN-094 | Voynich | Nmi class length | 0.097 | Latin 0.094 · slot 0.044 · shuffled 0.054 | 22 Oct |
| RUN-093 | Voynich | Nmi class last letter | 0.156 | Latin 0.158 · slot 0.061 · shuffled 0.065 | 22 Oct |
| RUN-092 | Voynich | Nmi class first letter | 0.177 | Latin 0.152 · slot 0.084 · shuffled 0.083 | 22 Oct |
| RUN-091 | Voynich | Vocab | 331.0 | Latin 310.0 · slot 343.0 · shuffled 347.0 | 22 Oct |

## Applied to

![Voynich Manuscript](https://mysteriesdept.com/media/folio/78r)

**[Voynich Manuscript](https://mysteriesdept.com/mysteries/voynich-manuscript)3 datasets** Exploration split, Latin genre texts, ZL3b transcription

[Download module](https://mysteriesdept.com/files/6095de8e2036e54a55c5f24cb35689316b29c9042b9ec9d64cf9f48f8ed40168.py)

Category

Nlp / copiale-style context clustering

Version

1.0 · 25 Oct 2026

Licence

MIT

Language

Python

Cite

[BibTeX](https://mysteriesdept.com/methods/MTH08JFXDCR6J0N/distributional-word-classes-ppmi-svd-embeddings-k-means/cite.bib) · [APA](https://mysteriesdept.com/methods/MTH08JFXDCR6J0N/distributional-word-classes-ppmi-svd-embeddings-k-means/cite.txt)

Develop this method

Methods are versioned like code. Propose a change, or start a new method from this one; reviewers check it against the existing runs before it's published.

 [Start a new method from this](https://mysteriesdept.com/developers)[Propose a change](https://mysteriesdept.com/developers)[Run on a dataset](https://mysteriesdept.com/developers)

Authors

**[apollo](https://mysteriesdept.com/u/apollo)** Wrote version 1.0