Microsoft makes important algorithm behind search engine Bing open source

Get a free Techzine subscription!

Microsoft has made one of the algorithms from the core of its search engine Bing open source. In this way, the company hopes to help developers create faster and easier to navigate applications.

This is the Space Partition Tree And Graph (SPTAG) algorithm, which is now available under the MIT License. The algorithm is bundled in a library that also contains tools to help developers put the code in their projects, writes Silicon Angle.


SPTAG is the algorithm that allows Bing to display directly relevant search results, even if a user enters a job that cannot be processed by simply matching keywords with web pages. For example, if you fill in the sentence “largest lake in the United States”, you will see a panel with information about Lake Superior. That while only one word is shared between the two.

SPTAG makes all this possible by converting searches into data structures called vectors. A vector is basically a long sequence of numbers that can contain different types of information. These can be individual words, but also complete web pages.

The advantage of translating different records into a common number format is that they are easier to compare. The phrase “largest lake in the United States” will have similarities with the vector that Bing generates from the text of the Wikipedia page “List of largest lakes in the United States by area”. On that Wikipedia page, Lake Superior is at the top of the list.

Accelerate Searches

Bing groups the vectors that represent web content on the basis of similarities, to speed up searches. According to Microsoft, SPTAG enables Bing to search through millions of pieces of data in a few milliseconds. The search engine has access to a repository of over 150 billion vectors, which is constantly being expanded with new content from the Internet.

SPTAG is available via GitHub.

This news article was automatically translated from Dutch to give a head start. All news articles after September 1, 2019 are written in native English and NOT translated. All our background stories are written in native English as well. For more information read our launch article.