Twitter has recently launched a more personalized search options for its users to find latest and relevant news, images, videos and Tweets. This personalized search option will improve Twitter users search experience. This option is developed based on two major features: relevance-filtering of search results and the identification of relevant images and photos. Developers had to rewrite the Earlybird, Blender and search infrastructure to develop this option.
History begins in 2008, Twitter upgraded its search team from 3 to 15 engineers. They scaled its real time search engine in two orders of magnitude. When the task is completed and new search option is ready to go, developers replaced the old search infrastructure without major service interruptions.
Every second on average over 2,200 new Tweets are done by users. When something big happens this number can increase up to 3,4x. Most of the time users look for Tweets with which others are engaged or memorable Tweets. In new search, users find the results that are most relevant to him. In it, personalized and filtered search results are displayed to particular users.
Twitter have already gave the scope to personalize and choosing interesting accounts to follow and now with the personalize search option the fun increased. Twitter’s ranking function works to access social graph gather relationship between searcher and the author of the Tweet. Even you are a user with few or no account to follow you will still get benefitted from this personalized searching mechanisms.
Twitter’s this new search option gives the scope to search for entities in Tweets like images and videos. The mechanism is real difficult because same image or video can be tweeted many times with different keyword each time. The new search engine look up the images and videos for matching Tweets and and store them in a custom hash map. When users search it sorts the map and returns with best images and video.