AI_SUMMARIZE - Realtime Compute for Apache Flink - Alibaba Cloud Documentation Center

This topic describes how to use AI_SUMMARIZE to generate text summaries with a large AI model.

Limitations

This function requires Ververica Runtime (VVR) 11.4+.
The throughput of AI_SUMMARIZE operators is subject to the rate limits of Alibaba Cloud Model Studio. When the rate limits for a model are reached, the Flink job will be backpressured with AI_SUMMARIZE operators as the bottleneck. In some cases, timeout errors and job restarts may be triggered.

Syntax

AI_SUMMARIZE(
  MODEL => MODEL <MODEL NAME>, 
  INPUT => <INPUT COLUMN NAME>,
  MAX_LENGTH => <MAX LENGTH>
)

Input parameters

Parameter	Data type	Description
MODEL <MODEL NAME>	MODEL	The name of the registered model. For more information, see Model settings to register a model service. Note: The output type of this model must be VARIANT.
<INPUT COLUMN NAME>	STRING	The data for the model to analyze.
<MAX LENGTH>	INTEGER	The maximum length of the model output. Note: This input parameter must be a constant.

Outputs

Parameter	Data type	Description
summary	STRING	The summarized content.

Examples

Test data

description

What is Apache Flink?

Apache Flink is an open-source, distributed compute engine for processing real-time data streams and batch data. It excels at stateful computations, handling continuous data like website clicks, IoT sensor data, and stock trades with low latency, high throughput, and exactly-once semantics. Flink seamlessly supports both stream and batch processing.

Test statements

The following sample SQL commands use the Qwen-Plus model and the AI_SUMMARIZE function to summarize the input data.

CREATE TEMPORARY MODEL general_model
INPUT (`input` STRING)
OUTPUT (`content` VARIANT)
WITH (
    'provider' = 'openai-compat',
    'endpoint'='<YOUR ENDPOINT>',
    'apiKey' = '<YOUR KEY>',
    'model' = 'qwen-plus'
);

CREATE TEMPORARY VIEW infos(id, description)
AS VALUES (1, '
What is Flink?
Apache Flink is an open-source, distributed compute engine for processing real-time data streams and batch data. It excels at stateful computations, handling continuous data like website clicks, IoT sensor data, and stock trades with low latency, high throughput, and exactly-once semantics. Flink seamlessly supports both stream and batch processing.
');

-- Use positional argument to call AI_SUMMARIZE
SELECT id, summary 
FROM infos, LATERAL TABLE(
  AI_SUMMARIZE(MODEL general_model, description, 10));
-- Use named argument to call AI_SUMMARIZE
SELECT id, summary 
FROM infos, LATERAL TABLE(
  AI_SUMMARIZE(
    MODEL => MODEL general_model, 
    INPUT => description, 
    MAX_LENGTH => 10));

Output result

id	summary
1	Apache Flink is a real-time stream processing engine.