tfmorris
commented
5 days ago I think this would be better done based on browser locale rather than language (which may have been what was meant).
I think Wikidata is the only reasonable source for this data in the general case, but, now that #7264 is resolved, we at least have the native language/script version along with the transliterated name (transliterated according to the cataloging rules of the institution that cataloged it).
A good first step would be importing all the name variants from Wikidata and adding them to the author AKAs to improve findability.
I think, it would be better if we showed translation of author name depending on language you selected rather than just showing same name for every user. Some people may not know how to read certain scripts like Cyrillic, Greek alphabet or Chinese characters and would need to open author page and look at alternative name section which is often cluttered. And it would be more convenient if patron could see author name by default in his language
This would require a lot of efforts (maybe, export from Wikidata?), but I think this is a better solution than Romanization.
If we automatically Romanize names, there would be a lot of mistakes. For example, Cyrillic character И is pronounced differently in Ukrainian and russian and therefore is differently Romanized (Y for Ukrainian, I (i) for russian) and we have no way to detect in which language author name is written. Though, maybe, we can use Wikidata to detect country of origin or nationality and only then Romanize? Another case is Chinese and Japanese. They have a lot of common characters because Japanese writing comes from Chinese, but they have completely different reading