Run · Voynich Manuscript

Topic model (LDA) of pages, k by held-out perplexity and fixed k=10: topic NMI with section / Currier language / scribe, half C, mean of…

Topic model (LDA) of pages, k by held-out perplexity and fixed k=10: Topic NMI with section / Currier language / scribe, half C, mean of five seeds.

Exploratoryraw 0.37 / 0.53 / 0.54; normalised 0.36 / 0.43 / 0.420

Result

Run	Measured	Result	Against	Decision	Date
RUN-257	Topic NMI with section / Currier language / scribe, half C, mean of five seeds	raw 0.37 / 0.53 / 0.54; normalised 0.36 / 0.43 / 0.42	Seed sd 0.04-0.08	Exploratory	28 Oct

What it used and found

Topic model (LDA) of pages, k by held-out perplexity and fixed k=10

Method

MTH

Exploration split

Dataset

DAT

Line-edge rules learned on half E (46 minimal alternations at paragraph start, line start and line end)

Dataset

DAT

ZL3b transcription

Dataset

DAT

Without the layout layer, topics follow scribe less

Evidence it produced

EVD

Parameters

half C, seeds 0-4

Cite it

Mysteries Department. (2026). Topic model (LDA) of pages, k by held-out perplexity and fixed k=10: topic NMI with section / Currier language / scribe, half C, mean of… [Run RUN0DCABMD05G5R]. https://mysteriesdept.com/r/RUN0DCABMD05G5R

@misc{mysteriesdept_run0dcabmd05g5r,
  title     = {Topic model (LDA) of pages, k by held-out perplexity and fixed k=10: topic NMI with section / Currier language / scribe, half C, mean of…},
  author    = {{Mysteries Department}},
  year      = {2026},
  note      = {Run, catalog key RUN0DCABMD05G5R},
  url       = {https://mysteriesdept.com/r/RUN0DCABMD05G5R}
}