How I improved Twitter Handles Search's ranking

[redox_]

http://blog.algolia.com/improving-ranking-twitter-handles-search/

[shawabawa3]

Seems pretty good, however if you search for "bill", the 3rd result is "Ryan Seacrest"

If there's only a partial match with the description it probably shouldn't rank it so highly

[redox_ → shawabawa3]

You're right "Ryan Seacrest" has +10M followers.

As you might have read on the blog, the user query produces 2 underlying queries. The first targeting top users (having 10M+ followers). This one retrieves "Ryan Seacrest", even if the match was only part of the description. It would be better if replaced by a query targeting the user's followers, but I reached Twitter's rate limits while implementing it. The second targeting others.

[redox_ → redox_]

I've just committed a patch skipping top users if the match was only in the description, with typos. Thanks for your feedback.

[redox_]

Demo is available at http://twittersearch.algolia.io