PAI-Rec developed in the Go programming language is an engine used for online recommendation systems. You can use this engine framework to quickly deploy recommendation services and perform custom development.
The PAI-Rec engine provides the following capabilities:
Integrates Go-based HTTP servers and supports router registering to enable RESTful API development.
Covers a complete pipeline for a recommendation engine. The pipeline predefines multiple types of recalling, filtering, and ranking policies, and integrates a module that can access Elastic Algorithm Service (EAS) of Platform for AI (PAI).
Loads multiple types of data sources, such as Hologres, ApsaraDB for Redis, Tablestore, and ApsaraMQ for Kafka.
Describes a recommendation process based on flexible configurations.
Integrates the lightweight A/B testing platform.
Supports FeatureStore that allows you to better manage features and data consistency.
Supports simple and easy-to-use secondary development to enable custom operations.
Overall architecture
System architecture
Flowchart of the recommendation service
Built-in features
