Jaa


Intelligent Recommendations architecture

Intelligent Recommendations is a hyper-extensible and scalable headless Microsoft Azure service that's easy to onboard and start using with zero-code integration.

Service flow

Intelligent Recommendations has two integration points:

  • Reading customer data on the back end using Microsoft Azure Data Lake Storage
  • Front-end serving experience that showcases recommendations results to the client side via secure HTTPS endpoints

All relevant data for Intelligent Recommendations machine learning, such as item catalog, availabilities, and other metadata and interactions (transactions), are structured in the Data Lake Storage file system and shared securely.

On the other side of the service, the client app calls the Intelligent Recommendations API to get the personalized or recommended content.

For more information about Data Lake Storage, go to Introduction to Azure Data Lake Storage Gen2.

Conceptual overview of Intelligent Recommendations architecture.

Important

Intelligent Recommendations doesn't store or process customer data outside the region you deploy the service instance in.

Step 1: Bring your own data

Intelligent Recommendations doesn't have a prerequisite license. Our compliant solution ensures that all company data stays within a defined customer geo. You can connect it to your company’s Data Lake Storage account.

Data type Description
Catalog General information about items, content, and other generic services Intelligent Recommendations recommend.
User interactions Interactions between users and items that Intelligent Recommendations models learn from and use to predict future interactions. Examples of user interactions include click streams, purchases, downloads, likes, and views.

Step 2: Run AI-ML service

After data is structured and shared, and the Intelligent Recommendations service instance is initiated, the "cooking" process begins. Data is processed and modeled according to business needs and scenarios. You can monitor progress by examining the output logs to make sure everything runs smoothly.

With an extensible architecture, businesses have the power to introduce more business logic and manage multiple instances of recommendations models. These multiple instances are useful for experimentation or for creating use cases with different signals.

Step 3: Call APIs to use results anywhere

Our solution integrates well on omnichannel platforms, using a simple recommendations API to create extensible, customizable experiences.

This solution provides real-time filtering and refreshed ordering of item results, and personalization of any list.

For examples, go to Example supported scenarios. For API information, go to Intelligent Recommendations API.

Try for free

You can try the Intelligent Recommendations free for three months with a one model, one RPS account. For more information, see QuickStart Guide.

See also

Deploy Intelligent Recommendations
Use data contracts to share data
Intelligent Recommendations API