Another key area of the artificial intelligence (AI)—the search also received several major updates that Microsoft announced at the Build 2018 conferenc in Seattle.
These six new advanced Bing Artificial Intelligence (AI) capabilities for Microsoft Cognitive Services enable innovative ways for users to interact with businesses and help them get rich, contextually-aware information.
Here are the deatils of new innovations unveiled at Build:
Bing Visual Search is now generally availabile let users identify entities and text within images and provide insights similiar to the image detils shown on Bing.com/images, so they can find sources for similiar images, products and objects. The searchable categories include fashion, landmarks, flowers, celebrities and more.
For example, just upload a picture for search insights about the image such as visually similar images, shopping sources, webpages that include the image, and more. Instead of upload, a URL of the image along with the optional crop box can guide visual search.
Also, if an image contains celebrities or a fashion object, the results will now include bounding boxes for regions of interest in the image.
Screen shot of similar image results with Bing Visual Search:
- Bing Visual Search now also understand text to scan business cards and extract structured information.
- Associate images with billions of webpages and provide image-based search from your domain.
Developers can now leverage the alpha release of the new Visual Search Developer Platform, to build skills on the platform and test them on the Bing iOS app. Microsoft said, they’ll make these skills available across Bing and Microsoft properties with Bing Visual Search included soon.
Watch the video below to see new Bing visual search experience and skills:
A request to Bing Visual Search API delivers one or more tags that contain the insights, such as webpages with image, visually similar images, and shopping sources for items found in the image.
Microsoft also deliver new functionalities to its ad-free search solution, the Bing Custom Search, like support to search across images on a domain, custom autosuggest witth type-ahead capability for search suggestions relevant to a domain.
This release also let you quickly test deployments with select individuals in your organization using the instance staging feature. Also, it now provide usage insights with a variety of metrics along with rich slicing and dicing capabilities.
Other capabilities announced today, include:
Bing Statistics add-in that allow you to derive insights from the metrics, such as call volume, top queries, API response code distribution, market distribution amongst many others is generally available now. “The rich slicing and dicing capability equips you to gather deeper understanding of your users and their usage to help inform your business strategy,” stated Bing.
Bing SDKs reached general availability are now available in programming languages, such as C#, Java, Node.js, and Python for development.
Upate 05/09: Microsoft released a private preview of two new Bing Maps APIs – Bing Maps Location Recognition API and Bing Maps Autosuggest API. The tow APIs will be generally available this summer, said Bing.
With Bing Maps Location Recognition, developers can fully recognize user locations, apps & services to build next gen rich experiences and analytics, like store specific notifications or coupons, visit tracking and identify new customer segments based on the types of location users visit.
The API response consists of:
- Business entities situated at the location. A wide variety of entity types are supported (e.g. restaurants, hotels, parks, gym, shopping malls and more).
- Natural entities at the location (e.g. beaches, island, lakes and 9 other types).
- Reverse geocoded address of the input location, with neighborhood and street intersection information where available.
- Type of property (e.g. residential, commercial) situated at the location
Bing Maps Autosuggest, when given a partial query, the service returns a list of suggested entities the user is most likely searching for, such as “roads, addresses, places and businesses (Businesses are currently constrained to en-US).”