Datasetpublic

dh-unibe/image-text_parlamentsdienste-protokolle

Dataset Card for transkribus-exports-280089-1-raw-xml This dataset was created using pagexml-hf converter from Transkribus PageXML data. Dataset Summary This dataset contains 479 samples across 1 split(s). Geographical scope: SwitzerlandPeriod: 1848-1900Languages: German, FrenchType of document: ProtocolsProvenance: Swiss Federal Archives Projects Included 1849_01 1849_02 1849_03 1849_04 1849_05 1849_06 1849_07 1849_08… See the full description on the dataset page: https://huggingface.co/datasets/dh-unibe/image-text_parlamentsdienste-protokolle.

sourceHugging Facemitupdated 6mo agoView on Hugging Face

0likes25downloads

Dataset Card

Dataset Card for transkribus-exports-280089-1-raw-xml

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

Dataset Summary

This dataset contains 479 samples across 1 split(s).

Geographical scope: Switzerland<br>Period: 1848-1900<br>Languages: German, French<br>Type of document: Protocols<br>Provenance: Swiss Federal Archives

Projects Included

—1849_01
—1849_02
—1849_03
—1849_04
—1849_05
—1849_06
—1849_07
—1849_08
—637157213248534270_verkleinert

Dataset Structure

Data Splits

—train: 479 samples

Dataset Size

—Approximate total size: 13347.73 MB
—Total samples: 479

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

python

from datasets import load_dataset

# Load entire dataset
dataset = load_dataset("dh-unibe/transkribus-exports-280089-1-raw-xml") 

# Load specific split
train_dataset = load_dataset("dh-unibe/transkribus-exports-280089-1-raw-xml", split="train")