Microsoft open-sources a crucial algorithm behind its Bing Search services

Microsoft today announced that it has open-sourced a key piece of what makes its Bing search services able to quickly return search results to its users. By making this technology open, the company hopes that developers will be able to build similar experiences for their users in other domains where users search through vast data […]

Microsoft today announced that it has open-sourced a key piece of what makes its Bing search services able to quickly return search results to its users. By making this technology open, the company hopes that developers will be able to build similar experiences for their users in other domains where users search through vast data troves, including in retail, though in this age of abundant data, chances are developers will find plenty of other enterprise and consumer use cases, too.

The piece of software the company open-sourced today is a library Microsoft developed to make better use of all the data it collected and AI models it built for Bing .

“Only a few years ago, web search was simple. Users typed a few words and waded through pages of results,” the company notes in today’s announcement. “Today, those same users may instead snap a picture on a phone and drop it into a search box or use an intelligent assistant to ask a question without physically touching a device at all. They may also type a question and expect an actual reply, not a list of pages with likely answers.”

With the Space Partition Tree and Graph (SPTAG) algorithm that is at the core of the open-sourced Python library, Microsoft is able to search through billions of pieces of information in milliseconds.

Vector search itself isn’t a new idea, of course. What Microsoft has done, though, is apply this concept to working with deep learning models. First, the team takes a pre-trained model and encodes that data into vectors, where every vector represents a word or pixel. Using the new SPTAG library, it then generates a vector index. As queries come in, the deep learning model translates that text or image into a vector and the library finds the most related vectors in that index.

“With Bing search, the vectorizing effort has extended to over 150 billion pieces of data indexed by the search engine to bring improvement over traditional keyword matching,” Microsoft says. “These include single words, characters, web page snippets, full queries and other media. Once a user searches, Bing can scan the indexed vectors and deliver the best match.”

The library is now available under the MIT license and provides all of the tools to build and search these distributed vector indexes. You can find more details about how to get started with using this library — as well as application samples — here.

Data & News supplied by www.cloudquote.io
Stock quotes supplied by Barchart
Quotes delayed at least 20 minutes.
By accessing this page, you agree to the following
Privacy Policy and Terms and Conditions.