Hi Ravi,

Thanks for writing about this reverse use case for our new language search detection, it is an interesting point.

I'm not sure that we'd want to automatically add in the English wikidata links to all the language wiki's because I'm not sure that those users would want that, especially in the wiki's that use the Latin alphabet.

For instance - on the Tamil Wikipedia, would users search for 'wolf' or are they really looking for a Latin transliteration of the Tamil word for wolf? If so, it would mean that the search would be trying to phonetically match English queries to non-English articles. For names - this transliteration is often the same, but for other things they're generally not the same at all.

I've cc'd the Discovery email list to see if there are additional thoughts and feedback on this question.

Cheers,

Deb

-- Deb Tankersley Product Manager, Discovery IRC: debt Wikimedia Foundation

On Wed, Aug 17, 2016 at 5:46 AM, Ravi ravishankar.ayyakkannu@gmail.com wrote:

Hi,

Regarding https://blog.wikimedia.org/2016/07/27/wikipedia-language-search :

There is actually a need for a reverse use case. If someone searches in English in Tamil Wikipedia, it should detect the correct Tamil article. This can be implemented using Wikidata links. Many non-roman script (especially, indic language users) language Wikipedia users search in English inside the search box. They actually hope to see an article in their own language. Just that they could not type in their own language. There are some Wikipedias which actually create 1000s of redirects with English titles to solve this problem.

Ravi