What does this Amazon Web Services Solution do?
Building a recommender systems is a challenging task as it requires specialized expertise in analytics, machine learning, software engineering, and systems operations, etc. Amazon Web Services provides an end-to-end, cloud-native recommendation system solution. The solution leveraging services such as
Amazon SageMaker,
Amazon Elastic Kubernetes Service(Amazon EKS),
Amazon CodeBuild, etc, allows you to significantly shorten the time to design and build your own recommender system.
The solution provides different recommendation algorithms, including the built-in algorithms of
Amazon Personalize
. You can easily switch between different recommendation methods, compare the recommendation results and choose the most suitable one for your business.
The solution is based on Microservice Architecture, which provides you the flexibility of building recommender system for various scenarios (for example, news recommendation, movie recommendation, course recommendation, etc.) .
Amazon Web Services Solution Overview
The diagram below presents the architecture of the solution. You can contact sales for deployment materials and get supports for customization.
Architecture Description
The solution will build a cloud-native recommender system for you from three aspects:
- Online service module. The solution leverages Amazon EKS to provide a set of microservices including user portrait, data loader, event notification, recall, rank, filter and so on. These services will interact with Amazon Redis.
- Offline processing module. The solution leverages Amazon Step Functions and Amazon SageMaker to implement the offline processing process, which includes data pre-processing, model training, model validation, batch processing and so on. Whenever new files or models are generated, the solution will notice the online services reload them.
- Automated release pipeline. The solution leverages Amazon CodeBuild and Argo CD to provide the CI/CD, which enables you to easily deploy the latest software into the production environment.
This solution also supports three recommendation methods of
Amazon Personalize, which can be respectively used in services such as recall, sorting, and events. The customer can easily switch between different recommendation methods.
Recommender System Solution
Version 2.0.0
Last updated: 09/2021
Author: Amazon Web Services
Features
Plugins
This solution offers built-in scenarios for recommendation (news and movie). The core services are designed as plugins enabling you to customize for other scenarios.
Scalability
Amazon Elastic Kubernetes Service (Amazon EKS) gives you the flexibility to start, run, and scale Kubernetes applications in the cloud.
Multiple Recommendation methods for your choice
This solution provides in total four recommendation methods, each of which contains different algorithm models and logics. You can switch to different methods, compare their recommendation results, and find the best method for your own business logic.
End-to-End
The solution contains all the basic components of a recommender system, allowing you to quickly build a complete recommender system. This solution also integrates the automated CI/CD pipeline which makes it easy for developers to release new implementations.
Explore all Amazon Web Services Solutions
Browse our portfolio of Amazon Web Services-built solutions to common architectural problems.
Find a Partner
Find Amazon Web Services certified consulting and technology partners to help you get started.
