Blog · Notes from the lab
Notes on scripts, data and the work in between.
Project notes, methods and research from the team: how we digitise, catalogue and publish multilingual heritage, and what we learn along the way.
Latest
-
DM-LP: a large Arabic-script dataset for OCR and cataloguing
Why we built a dataset of Arabic-script title pages, and how it serves both OCR research and library cataloguing.
-
Transliterating Arabic for library catalogues
REICAT, ISO 233-2 and the small decisions that make a bilingual MARC 21 record searchable on both sides.
-
FAIR data for religious studies: lessons from RESILIENCE
Mapping partner datasets to schema.org JSON-LD, and the documentation that makes them reusable.
-
Mem at the Rome AI Hub Ecosystem Event
On 3 November Mem takes part in the Rome AI Hub Ecosystem Event at Luiss, in Rome.
-
When OCR loops: repetitive degeneration in neural text recognition
Notes on a failure mode of neural OCR that every pipeline on non-Latin scripts eventually meets.
No posts in this category yet.
Example posts for the mock-up: topics drawn from our real work, texts and dates indicative.
One email when something new is out.
No more than once a month. Unsubscribe at any time.