dh-unibe/image-text_aaeb-xiv-xvii-part-2

Dataset Card for transkribus-exports-127147-raw-xml

This dataset was created using pagexml-hf converter from Transkribus PageXML data.

Dataset Summary

This dataset contains 121 samples across 1 split(s). This dataset contains images and transcription from the Archives de l’ancien Evêché de Bâle.

Most texts are in Latin, French, and Early Modern German.

Geographical scope: Switzerland<br>Period: 1400-1500<br>Languages: Early Modern German<br>Type of document: Protocols<br>Provenance: Archives de l'ancien Evêché de Bâle<br>

Projects Included

• —B16814-5
• —B16815-10_2
• —B16815-10_3
• —B16815-103testEP
• —B16815-20_2
• —B16815-23
• —B16815-2_1
• —B16815-2_2
• —B16815-2_3
• —B16815-2_4
• —B16815-3
• —B16815-5_4
• —B16815-8_17
• —B16815-9_2
• —B16815-9_3
• —B16815-9_4
• —B16817-1_2
• —B16817-1_3
• —B16817-1_4
• —B16817-1_5
• —B16817-1_6

Dataset Structure

Data Splits

• —train: 121 samples

Dataset Size

• —Approximate total size: 1430.95 MB
• —Total samples: 121

Features

• —image: Image(mode=None, decode=False)
• —xml_content: Value('string')
• —filename: Value('string')
• —project_name: Value('string')

Data Organization

Data is organized as parquet shards by split and project:

data/
├── <split>/
│   └── <project_name>/
│       └── <timestamp>-<shard>.parquet

The HuggingFace Hub automatically merges all parquet files when loading the dataset.

Usage

from datasets import load_dataset

# Load entire dataset
dataset = load_dataset("dh-unibe/transkribus-exports-127147-raw-xml") 

# Load specific split
train_dataset = load_dataset("dh-unibe/transkribus-exports-127147-raw-xml", split="train")