Macronizer/oga-macronized
OGA Macronized Automatically macronized Ancient Greek text: the phonemic length of the three "dichrona" (ambiguous-length letters) — alpha (α), iota (ι), upsilon (υ) — has been marked wherever the grc-macronizer rule-based system could disambiguate it with confidence. A trailing ^ marks a short dichronon, a trailing _ marks a long one; dichrona the macronizer could not resolve are left unmarked (never guessed). Source Text is the Opera Graeca Adnotata corpus… See the full description on the dataset page: https://huggingface.co/datasets/Macronizer/oga-macronized.
054
1---2language:3- grc4license: gpl-3.05task_categories:6- text-generation7- token-classification8tags:9- ancient-greek10- vowel-length11- macronization12- classics13pretty_name: OGA Macronized (Ancient Greek Vowel Length)14size_categories:15- 1M<n<10M16---17 18# OGA Macronized19 20Automatically macronized Ancient Greek text: the phonemic length of the three21"dichrona" (ambiguous-length letters) — alpha (α), iota (ι), upsilon (υ) — has22been marked wherever the grc-macronizer rule-based system could disambiguate it with confidence. A trailing `^` marks a short dichronon, a trailing `_` marks a long one; dichrona the macronizer could not resolve are left unmarked (never guessed).23 24## Source25 26Text is the [Opera Graeca Adnotata](https://doi.org/10.5281/zenodo.14206061)27corpus (Celano 2024) — ~40M tokens of Ancient Greek aggregating First1KGreek,28the Patristic Text Archive, and Perseus' canonical-greekLit, with lemma/POS/29morphology already annotated by Celano's own parser. That existing annotation30was converted directly into the macronizer's internal representation (bypassing31any separate NLP pipeline) and run through the rule-based macronizer.32 33- 2,234,599 sentences34- Overall dichrona-in-open-syllables macronization ratio: 69.20%35 (11,627,591 / 16,802,080 disambiguated)36 37## Fields38 39- `macronized`: the sentence with `^`/`_` markup, e.g. `χρὴ γι^νώσκειν ὅτι^ πά_σης τῆς γῆς`40- `plain`: the same sentence with markup stripped (i.e. ordinary Greek orthography)41 42## Intended use43 44Training data for sequence models that predict Ancient Greek vowel length45(e.g. a char-level tagger), and as an intermediate resource for downstream46prosody-aware NLP (verse scansion, metrical classification, morphology47disambiguation of the dichrona-bearing minimal pairs like ἄν/ἂν).48 49Note: labels are *silver*, not gold — they come from a rule-based system that50is deliberately conservative (does not guess), not from human annotation.51 52## License53 54The underlying OGA corpus is distributed by its authors under the terms shown55on its [Zenodo record](https://doi.org/10.5281/zenodo.14206061); check that56record for the authoritative license before redistribution. The macronization57itself is produced by software licensed under GNU GPL v3.58 