Datasetpublic

Urdatorn/oga-conllu-stoicheia

OGA parsed with Stoicheia Period-delimited sentences from the source=oga records of the pristine split of Ericu950/AncientGreek. A literal . terminates a sentence and is retained. All source columns are inherited by each sentence. author contains the canonical author label corresponding to the TLG/CTS code in id; text contains the sentence and conllu contains the Universal Dependencies-style output from Ericu950/Stoicheia-tagger-parser. The dataset contains 1,135,428 sentences… See the full description on the dataset page: https://huggingface.co/datasets/Urdatorn/oga-conllu-stoicheia.

sourceHugging Facecc-by-sa-4.0updated 2mo agoView on Hugging Face

0likes257downloads

Dataset Card

OGA parsed with Stoicheia

Period-delimited sentences from the source=oga records of the pristine split of Ericu950/AncientGreek. A literal . terminates a sentence and is retained. All source columns are inherited by each sentence. author contains the canonical author label corresponding to the TLG/CTS code in id; text contains the sentence and conllu contains the Universal Dependencies-style output from Ericu950/Stoicheia-tagger-parser.

The dataset contains 1,135,428 sentences in 114 Parquet shards. Columns are ordered as follows:

Column	Description
`author`	Canonical OGA/TLG author name, for example `Sophocles`
`source`	Source corpus identifier
`id`	CTS text identifier containing the TLG author and work codes
`license`	Source record license
`tier`	Source dataset tier
`orig_band`	Original source band
`clean`	Source cleanliness score
`text`	Period-delimited sentence text
`conllu`	Stoicheia Universal Dependencies-style CoNLL-U annotation

Canonical author labels come from the OGA text catalogue.