Urfehdenbuch X: Handwritten Text Ground Truth for Basel (1563-1569)

Ground truth data pairs images of historical handwritten pages from Basel's "Urfehdenbuch X" with their aligned transcriptions. The dataset provides Page XML and ALTO files for text-to-image alignment, derived from a digital edition project. Tobias Hodel from the University of Bern prepared this set for Handwritten Text Recognition (HTR) research.

Use Cases

• Train Handwritten Text Recognition models based on the aligned image-text pairs.
• Benchmark OCR/HTR algorithm performance on historical German script from the 16th century.
• Analyze historical legal language and record-keeping practices based on the transcribed oaths ('Urfehden').
• Develop document layout analysis tools using the provided ALTO and Page XML structure files.

Strengths

• Provides professionally aligned ground truth between page images and transcriptions.
• Includes structured text data in both ALTO and Page XML formats, which likely contain layout and semantic information.
• Derived from a cited scholarly digital edition project, suggesting curation by domain experts.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Staatsarchiv Basel-Stadt (StABS), Ratsbücher O10, Urfehdenbuch X.

Collection Method

Images and text aligned using text-to-image tools within Transkribus; TEI to text conversion by Peter Dängeli.

Time Range

1563-1569

Geography

Basel, Switzerland