The PAI-Rec engine is a Go-based framework for online recommendation services. You can use this framework to quickly build online recommendation services or perform custom development.
Basic features
The PAI-Rec engine provides the following features:
Integrates a Go HTTP server and provides route registration to simplify RESTful API development.
Includes a complete pipeline for the recommendation engine with predefined retrieval, filtering, and sorting policies, and provides built-in access to Alibaba Cloud Elastic Algorithm Service (EAS).
Loads data from multiple data sources, such as FeatureDB, Hologres, Redis, Tablestore, and DataHub/Kafka.
Uses flexible configurations to describe the recommendation flow.
Integrates a lightweight A/B testing platform.
Supports FeatureStore to improve feature management and ensure data consistency.
Supports easy-to-use extension points for custom operations.
Overall architecture
The overall architecture of the PAI-Rec engine is as follows:
System architecture and deployment
Overall flow of the recommendation service
Built-in features
