WORK · CASE 01
A canonical metadata contract for Arabic books
One schema that serves ingestion, discovery and distribution, built for a language the standards forgot.
The problem
Arabic titles reached Samawy with inconsistent, incomplete and sometimes duplicated metadata. Global channels need clean ONIX, subject codes and descriptions. Existing subject schemes were built for other languages.
What I did
I defined a canonical metadata contract: a 94-column schema with a 47-section addendum. I built rules that collapse the Thema taxonomy into a 37-category interface taxonomy in Arabic, and set up the mapping from BISAC to Thema, Dewey and ONIX 3.0. I also designed a bilingual enrichment step that fills only missing fields, using content extracted from the book, with a confidence score for each field.
What it produced
One contract that serves ingestion, discovery and distribution. Duplicates are treated as a defect, not a cleanup task. The work continues as part of the platform roadmap.