News

Microsoft Open Sources Bing Search’s Most Advanced Algorithm

Microsoft has announced that it is open-sourcing a crucial AI tool behind its Bing search engine. This tool enables Bing to quickly return relevant search results to user queries.

The company has open-sourced its AI algorithm called Space Partition Tree And Graph (SPTAG) that can parse data more efficiently.

Earlier, performing a web search was simple. Users typed a few words and went through pages of results. In the present day, with the advent of techs like Google Lens and Bing Visual Search, users can take a picture instead and drop it into a search box to get search results.

Or other times, they use an AI assistant to ask a question and have them reply without physically touching a device at all. Many users simply type a question and expect an actual reply and not a list of pages with probable answers.

This is where the Space Partition Tree And Graph steps in.

Microsoft explains in its blog post that the SPTAG helps developers in sifting through the data through vectors (mathematical representations of words, image pixels, and other data points) in milliseconds.


For the uninitiated, SPTAG is written in C++ language and is at the core of the open-source Python library. It is the most important pillar for a number of Bing Search services and Microsoft says that it helps the company “better understand the intent” behind the millions of web searches performed each day.

For instance, typing “How tall is the tower in Paris?” in Bing gives you the right answer — 1,063 feet — even though you never mentioned the word “Eiffel” in the question and “tall”  never appears in the result.

The aim behind making this technology available to all is to help developers create a similar experience for users when they search on other platforms where there are huge amounts of data such as retail.

You can find the open-sourced code for SPTAG at GitHub and watch how Microsoft uses SPATAG algorithm in Bing in this video:

Click to comment

You must be logged in to post a comment Login

Leave a Reply

To Top

Pin It on Pinterest

Share This