Support multiple languages in CSV representations of ontologies

[jvendetti]

The CSV representations that we generate for ontology submissions only include English content for preferred labels — the writer selects the English-tagged label, falling back to an untagged label, then the first available (see #288 / #292). Now that multilingual ontology support has been introduced in BioPortal, we've received a request from an end user to extend the CSV representations to include preferred labels in additional languages.

The request concerns the RadLex ontology, currently available in English and German:

... I know I said that including the German preferred term as well is not as important as getting the English preferred terms. But of course now we have a German who is interested in reviewing all the translations for us and asks for a spreadsheet with both. The new one does have the German synonyms, but not the preferred terms. I don't know how much trouble it would be to add those as well, but if it wouldn't be too much extra, we would love to have those as well as a separate column.

As the user notes, the current CSV output already includes German synonyms (synonym values are written as-is, without language filtering), while preferred labels are limited to English.

Two approaches have been suggested so far:

1. Per-language columns — what the end user requested: keep the existing Preferred Label column (English-first, backward compatible) and add separate columns for other languages, e.g., Preferred Label (de).
2. Language-tagged values in existing columns — suggested in CSV Format Change: prefLabel Field Now Contains List Instead of String Value #218 (comment): keep the columns singular, but have each cell contain multiple values with language tags attached (e.g., colon @en|kolon @de), applied uniformly to all multilingual fields including synonyms. This would not be backward compatible.

Open design questions:

• Which of the two approaches above to adopt (or a hybrid)
• Whether the change should extend beyond preferred labels to synonyms, definitions, and other annotation properties
• Whether to include all languages present in the data or only declared submission languages
• How much weight to give backward compatibility for existing CSV consumers