andrawaag
commented
5 years ago @andrewsu That snapshot was made in this collab notebook.
The sparql query used there is here. The query only lists translations for diseases that do have a Wikipedia article.
There are however more translations available if we consider all Wikidata items. This SPARQL query returns those.
andrewsu
The basic premise is that Wikidata has lots of translations to other languages, and we (and others) have also loaded many ontologies. The natural extension is to use Wikidata as a backbone for translating ontologies to other languages. There are a few potential components here:
Download all existing translations for a given ontology. @andrawaag has already done this for Disease Ontology and posted an snapshot here: https://suwulab.slack.com/archives/C02G04Q6E/p1538134542000100 (Andra, is this the output of a raw sparql query? If so, can you post here?)
Create a web interface to generalize and automate query/download. Basically, users would select a language, an ontology, and an output format (CSV / OBO / OWL / ROBOTS). Optional parameter includes filtering to only include translations that agree with Google Translate.
Amend web interface to allow contributions of translations. Users would still select a language and an ontology, and the web interface would show terms for which no translation existed. Can show Google Translate translation as starting point. User can add a translation or skip to the next one. If users add translation, it could be added directly via OAuth (similar to Mix n Match), or can be stored in a local DB for aggregation and consistency checking.
tagging @andrawaag @stuppie