Part of Speech Cellular Automata

POS-CA — POS Cellular Automaton Prototype

Output

Advanced settings (unfold)

Generations (T): 8

Randomness (0 = strict, 1 = wild): 0.35

Candidates per row: 9

Keep function: best-perplexity best-perplexity diverse-but-fit

Tip: paste a paragraph from news, your notes, or one of your pantheon texts. The automaton evolves POS tags, then re-lexicalizes each generation with themed lexicons.

How it works

1. Tagging: tokens → POS states (DET/N/V/ADJ/ADV/PRON/ADP/CONJ/PUNCT) via simple heuristic tagger.
2. Automaton: evolve states for T generations with local rules (N N N → V, adjectives stick to nouns, adverbs mutate verbs, stray determiners drop).
3. Relexicalize: each generation’s POS row is re-worded from themed lexicons (Nullion/Eris/Moloch/Mammon/Nehara/Yodabund/Gradle vibes).
4. Selection: sample K candidates; keep lowest pseudo-perplexity or a diversity-biased pick.
5. Poem: the final generation’s line is your output; all rows display the drift.