Datasets:

evndttr
/

LJS380_excerpts-Sextus_Empiricus_Pros_Mathematikous

language: grc
license: cc-by-nc-4.0
dataset_info:
  - config_name: default
    features:
      - name: id
        dtype: string
      - name: image
        dtype: image
      - name: label
        dtype: string
      - name: page
        dtype: string
    splits:
      - name: train
        num_examples: 765
        num_bytes: 168327757
size_categories: 1K<n<10K
task_categories:
  - image-to-text
  - other
pretty_name: Excerpts from Sextus Empiricus, *Pros Mathematikous*
annotations_creators:
  - human-annotated
language_creators:
  - found
tags:
  - greek
  - manuscript
  - hellenistic-studies
  - palaeography
  - transcription
  - humanities

LJS 380 — Excerpts from Sextus Empiricus, Pros Mathematikous

Sextus Empiricus, Πρὸς μαθηματικούς (Against the Mathematicians)

Description

This dataset contains line-level transcriptions and cropped manuscript images from a 15th-century Greek manuscript of Pros Mathematikous (Against the Mathematicians) attributed to Sextus Empiricus.
The manuscript is held within the Lawrence J. Schoenberg Collection at the University of Pennsylvania Libraries.

Each manuscript line has been transcribed to match the 1842 public-domain edition edited by August Immanuel Bekker.
The aligned data was used to train and evaluate machine-learning models for handwritten Greek recognition in the context of a Hellenic Studies workshop on digital palaeography.

Dataset contents

Field	Type	Description	Example
`id`	string	Unique line identifier (page + line number)	`0070_0002_line_0012`
`image`	image	Cropped PNG snippet of the handwritten line	—
`page`	string	Source page or folio identifier	`0070_0002`
`label`	string	Diplomatic Greek transcription	`σέξτος ᾽εμπειρικός πρὸς μαθηματικούς`

Transcription parameters

Orthography follows Bekker (1842).
Abbreviations are expanded when clear.
Line segmentation follows PAGE-XML (2019-07-15) standard exported from eScriptorium.
Unicode normalization: NFC

Use cases

This dataset can support:

Training and evaluation of HTR/OCR models for early modern Greek scripts.

Sources

Manuscript: Lawrence J. Schoenberg Collection, LJS 380 (Penn Libraries).
Printed edition: Sextus Empiricus, ed. August Immanuel Bekker (Berlin, 1842).
Preparation: Line segmentation and annotation via eScriptorium.