WebProNews

Twitter Shares Some Search Ranking Details

As previously reported, Twitter has launched some new search-related changes for its iOS and Android apps.

Here’s what Twitter said about search in the announcement: “Search results now surface the most relevant mix of Tweets, photos, and accounts, all in one stream (similar to the stream in Discover). We’ve also added a new search button to Twitter for iPhone, letting you search from anywhere within the app. (This button was already available in the Android and iPad apps.) Look for the magnifying glass icon next to the button you use to compose a Tweet.”

In a separate blog post on the company’s engineering blog, Twitter talks a bit about how it ranks search results.

Youngin Shin of Twitter’s search quality team (Twitter has a search quality team) explains, “When a user searches, different types of content are searched separately, returning a sequence of candidate results for each content type with a type-specific score for each. For certain content types that are displayed as a single group or gallery unit, such as users or images, we assign the maximum score of results as the representative score of this content type. The result sequences for some content types may be trimmed or discarded entirely at this point.”

“Once results of different content types are prepared, each type-specific score is converted into a universally compatible score, called a ‘uniscore’,” adds Shin. “Uniscores of different modules are used as a means to blend content types as in a merge-sort, except for the penalization of content type transition. This is to avoid over-diversification of content types in the blended result.”

Twitter Search

As Shin explains, all pieces of individual content are assigned type-specific scores called “raw” scores by their corresponding services. Raw scores are then converted into uniscores using type-specific log-linear score conversion functions. According to Shin, the chance of a converted score taking its value in [0, 1] is at least 95%.

Read the full post here. It’s an interesting look into the back end of Twitter Search.