Challenged with rethinking how to build a movie recommendation experience, a team of Garage interns based out of Cambridge, MA created a sample app and corresponding documentation that shows how to use recommendation algorithms in an app experience. Today, we’re excited to share their project ahead of its debut at the RecSys’19 conference next week: Recommenders Engine Experience Layout, a Microsoft Garage project. This work joins a collection of best practices and tools for recommendation engines available on a larger Recommenders GitHub. Explore both on the Recommenders GitHub repository and Recommenders Engine Example Layout GitHub repository respectively.
Bringing recommendation tools to apps
Recommenders Engine Example Layout, focuses on recommendation algorithm experiences that take place in apps and provides a detailed breakdown of how developers can leverage the work by the sponsoring team. The Azure AI Customer Advisory Team, or AzureCAT AI works with such customers as ASOS to incorporate enhanced recommenders algorithms into their solutions. The team was inspired to partner with a team of Garage interns upon continued feedback that expanding on their popular Recommenders repository with a focus on apps would be helpful to customers who already have an app infrastructure.
“The key thing we wanted to demonstrate out of this was showing the recommenders we have, in a real-world setting that’s relevant to apps,” shares Scott Graham the Senior Data Scientist on Azure AI CAT who oversaw the project.”Often when we work with customers, they already have complex infrastructure and want to see how they can incorporate these algorithms into an app. This was a great opportunity to illustrate and document this.”
The Garage project documents how to build a sample app powered by the Recommenders algorithms, featuring the MovieLens dataset, one of the largest open source collections of movie ratings. Put by Bruce Gatete, Program Manager Intern for the project. “It provides an end-to-end demonstration of how developers can build fully function cross-platform applications that use these algorithms.”
Sample app key features
The sample app developers can build includes a wide variety of features, including:
- Browse a large dataset of movies
- Select and view movie descriptions
- Create your own personalized favorites list
- Switch between different recommender algorithms
- Switch between pre-generated personas or create your own
The Azure AI CAT team also has a continuous goal to accelerate the speed with which they’re able to partner with customers on this solution. Scott continues, “This was a great proof point that we could do this in the space of a summer timeframe: can we use these algorithms to quickly putt together an app? And we did!”
In addition to trying this project, developers can explore the original Recommenders repository which has recommendation algorithm tools and best practices such as a popular deep dive into the SAR model or an example of deploying these models in a production setting.
Built using Xamarin.Forms
The Recommenders Engine app is built using Xamarin.Forms and supports iOS, Android, and UWP platforms. “It was really great leveraging Xamarin Forms to be able to deploy this across all these platforms so quickly. I was really impressed with the speed of that coming together,” shared Scott Graham, who oversaw the project from the sponsoring team.
Try it Out
Become a Garage Intern. We’re hiring
The Garage is hiring for the 2020 Winter & Summer seasons! Here you can learn more details about the internship and how to apply.
Why become a Garage intern? The Garage opens doors to interesting and challenging projects and collaborative partners. Michelle shares her favorite part about the internship “I enjoyed getting to know my team very well over the summer–there’s an incredible amount of talent on our team, and it’s awesome getting to work with a team of people my age and truly have a say in the design and final result of our product.”