Twitter announced that now the search results in the service will be ranked by relevance, not time of publication.
Tests have shown that this approach leads to increased user engagement, and they spend more time in the service.
To determine the relevance of search results is used the algorithm based on machine learning, analyzing user behavior.
Before implementing this change the results of a Twitter search into several categories: popular, latest tweets, etc. But in each of them the posts are displayed in reverse chronological order.
Recall that in March this year, the news in Twitter has become an algorithmic default.
