Everything Delivered. No One Can Fully Process It.
TEI, IIIF, PAGE-XML and PROV-O each cover a slice. Why the real challenge at HistoriaMP begins only after analysis - in keeping scholarly status intact across system boundaries.
Read article
The blog collects articles on historical manuscripts, OCR/HTR, AI image analysis, model input, MUFI, Unicode, glyphs, minim clusters and source-bound reading. It is designed as a growing knowledge archive for Digital Humanities.
TEI, IIIF, PAGE-XML and PROV-O each cover a slice. Why the real challenge at HistoriaMP begins only after analysis - in keeping scholarly status intact across system boundaries.
What happens when AI-assisted manuscript analysis documents every intermediate step? Why full transparency creates a new paradox - and how HistoriaMP keeps traceability usable anyway.
Why HistoriaMP democratizes not truth itself, but the auditable path toward a reading: visual evidence, candidate status, uncertainty marking, review, AI and HTR as controlled tools.
How HistoriaMP uses Cappelli, Abbreviationes and MUFI as separate reference layers to create verifiable candidates from image findings and sign forms instead of smooth instant readings.
Why automatic text recognition and large language models are useful for historical manuscripts - but methodologically insufficient without visible evidence, uncertainty status and model-input documentation.
Why historical manuscripts need more than automatic text recognition - and why abbreviation signs, MUFI, Unicode, glyphs and minim clusters require documented uncertainty.
Why image scaling in AI analyses must be documented - and why an AI statement initially applies only to the concrete image version that was actually presented to the model.