DO Translation

Ontology Files

Using IRIs (aka PURLs) to access ontology files is strongly recommended. IRIs are persistent identifiers that will reliably access the latest version of the ontology files, regardless of where they are stored. The files available with their IRIs are:

• International version (all languages included): http://purl.obolibrary.org/obo/doid/translations/doid-international.owl
• Spanish-only versions
  • doid-es.owl (RDF/XML): http://purl.obolibrary.org/obo/doid/translations/doid-es.owl
  • doid-es.obo: http://purl.obolibrary.org/obo/doid/translations/doid-es.obo

Files can also be located on GitHub in the src/ontology/releases/translations directory of the HumanDiseaseOntology GitHub repository. Detailed information about accessing ontology files can be found in README-release-files.md.

Website Text

The first translation of website text was completed by professional medical translators at The Spanish Group prior to any translation of ontology terms. These translations were reviewed by fluent Spanish speakers on the DO team and found to be highly accurate except for ontology-related technical terminology, which medical translators would be unfamiliar with. As part of this review, the DO team established standards for named entities (like "Human Disease Ontology") and manually revised ontology-related terminology and explanatory text.

All subsequent translation of website text to Spanish is done immediately after updates to English (new or changed text) with Google Translate and then reviewed by fluent Spanish speakers on the DO team, or handled directly by a fluent speaker on the DO team (for minor changes). Prior translation trials demonstrated that Google Translate is also highly accurate for Spanish translation, including ontology-related text.

Website language switching is handled by the Python Babel library for text and server-side logic for OBO & OWL tree browsers and search.

Ontology Terms

A three-step protocol was used to ensure high quality translation of ontology text to Spanish.

1. Ontology term labels, definitions, and synonyms were translated by professional medical translators at The Spanish Group.
2. The same English text was translated to Spanish and professional Spanish translations were back translated to English using Google Translate and used to calculate a translation quality score, computed as the average similarity between the English-paired text and the Spanish-paired text.
3. Guided by translation quality scores, translated text was handled as follows:
   1. Exact matches across both languages were accepted without review.
   2. Comparisons scoring above a threshold were accepted, and will be minimally reviewed as time permits (0-1 scale, 1 best; threshold for labels & synonyms = 0.75, definitions = not yet determined).
   3. Comparisons scoring at or below the threshold are carefully reviewed by DO team members prior to inclusion into the ontology**.

**An appropriate threshold for definitions has not yet been determined. Until determined, professional medical translations of definitions will be included in the ontology without review.