Data source concepts
Currently, the Artificial Intelligence Recommendation system only supports MaxCompute (ODPS) as a data source. Support for other sources, such as OSS and RDS, will be added.
MaxCompute: MaxCompute, formerly known as ODPS, is a fast, fully managed data warehouse solution for TB/PB-scale data. For more information, see What is MaxCompute?.
Table: A table is a data storage unit in MaxCompute. It is a two-dimensional structure of rows and columns. Each row is a record, and each column is a field with a specific data type. A record can have one or more columns. The column names and data types make up the table schema. For the table schema that Artificial Intelligence Recommendation requires for full data, see the data specifications for Content industry, News industry, and E-commerce industry.
Partition: A partitioned table is a table for which you specify certain fields as partition key columns when you create the table. For more information, see Partition.
Project: A project is the basic organizational unit in MaxCompute. It is similar to a database or schema in a traditional database and acts as the primary boundary for multi-user isolation and access control.
Client: The client is the command line interface for MaxCompute. For more information, see Connect by using the client (odpscmd).
Full data preparation
1. Activate the MaxCompute service
If you already use the MaxCompute (formerly ODPS) service, you can skip this step. Otherwise, you must activate the service to upload full data to the Artificial Intelligence Recommendation system. For instructions about how to activate the service, see What is DataWorks?.
2. Upload full data to the data source
MaxCompute supports multiple data upload methods. If your data is already on Alibaba Cloud, such as in OSS or RDS, you can use the data integration feature in DataWorks. To upload data from a local machine to MaxCompute, see Common data upload scenarios and tools.
3. Grant permissions to a RAM user
The Artificial Intelligence Recommendation service requires permissions to sync full data from your MaxCompute project. Grant permissions to a RAM user and provide the AccessKey pair (AK) of that user. For detailed instructions, see MaxCompute offline storage authorization.
4. Create a data version in the Artificial Intelligence Recommendation console
For more information, see the "Start an instance" section in Start instance with MaxCompute historical data.
5. Subsequent operations for the data source
The Artificial Intelligence Recommendation service syncs the full data only once during initialization. Subsequent data is pushed incrementally using the SDK. The service does not use the data from the original data source after the initial sync. If you do not need to create a new data version or replace the full data, you can freeze or delete the service that you previously activated.