Arabic and Persian with one heart

A substantial portion of the school's heritage — advanced-seminary lecture transcripts (taqrirat), books of thought, and what was written around them — is in Persian and has not yet been translated. Confining the archive to the translated would truncate the system.

So we adopted the approach followed by multilingual European legal thesauri (the EUR-Lex/EuroVoc pattern): a concept is a single neutral node named in standard Arabic, and texts in their languages point to it. The Persian text enters the archive in its own script, its concepts are extracted under their unified Arabic names, and it meets its Arabic counterpart in the knowledge graph on the very same node.

And in the answer: you ask in your language and it answers in it; if the evidence is in Persian, the bot states: ‘what was stated in Persian, rendered into Arabic, is: …’ with the usual attribution, while the verbatim quotation remains in its original language, possibly followed by an indicative Arabic rendering outside the quotation marks.

And twelve languages as a conversation interface: ask in any of them and receive the explanation in it, while the approved text is relayed from its source.