Combine user interaction data with contextual data to generate high-quality recommendations
Using machine learning, Amazon Personalize can learn from past user interactions (events) such as clicks, purchases, watches etc. as well as information about the user such as age, location etc. and information about the item such as brand, price etc. to generate highly relevant recommendations for each user.
Automated machine learning
Amazon Personalize includes AutoML capabilities that take care of machine learning for you. Once you have provided your data via Amazon S3 or via real time integrations, Amazon Personalize can automatically load and inspect the data, select the right algorithms, train a model, provide accuracy metrics, and generate personalized predictions.
Based on the same technology used at Amazon.com
Amazon Personalize includes algorithms that are based on over twenty years of personalization experience and expertise developed from running the Amazon.com retail business. Amazon Personalize provides a range of algorithms suited for user personalization, similar item and reranking use cases.
Continuous learning to improve performance
Learn from every user interaction and continually improve your business objectives; Amazon Personalize allows you to send user events in real time and generate recommendations which respond to real time user activity. Customers can also retrain models on latest, up to date user events, user data and item data which enables Amazon Personalize to continuously calibrate to evolving user preferences.
Easily integrate with your existing tools
Amazon Personalize can be easily integrated into websites, mobile apps, or content management and email marketing systems, via a simple inference API call. The service lets you generate user recommendations, similar item recommendations and personalized reranking of items. You simply call the Amazon Personalize APIs and the service will output item recommendations or a reranked item list in a JSON format, which you can use in your application.
GetRecommendations API returns a list of relevant items given a userID. A representative usage example would be a content recommendation widget on landing page of a video streaming website that suggests a list of videos based on the user’s past watches.
GetRecommendations API can also be used to return a list of similar itemIDs given an input itemID. A representative use case is to recommend similar movies when a user is on the detail page of a movie.
GetPersonalizedRanking API reranks a list of itemIDs given a userID and a list of itemIDs to be reranked. The input list can be from any source, for example from an editorially curated list or from a list of itemIDs resulting from a search query. For example, an ecommerce retailer can use what they know about their customers’ previous behavior and past purchases to show the most relevant results, instead of showing the list of products that directly match the keyword.
GetRecommendation and GetPersonalizedRanking APIs can also be used for integration with existing email and notification workflows. For example, an online travel site can send a notification with the right promotional offer to a user if they abandon a flight search session on their app. By serving up relevant promotions and offers suited for each user, conversions and sales can be increased.
Refer to developer guide for instructions on using Amazon Personalize.
Instantly get access to the AWS Free Tier.
Get started building with Amazon Personalize in the AWS Console.