Amazon Personalize Pipeline

Manage resources of Amazon Personalize using API Gateway

Running this repository may cost you to provision AWS resources

Prerequisites

• awscli
• Nodejs 14+
• Python 3.10+
• AWS Account and Locally configured AWS credential

Installation

Install project dependencies

$ cd infra
$ npm i

Install cdk in global context and run cdk bootstrap if you did not initailize cdk yet.

$ npm i -g [email protected]
$ cdk bootstrap

Deploy CDK Stacks on AWS

$ cdk deploy "*" --require-approval never

API Specification

APIs provided this repository are self-documented using API Gateway Models.

1. Export Swagger your API specification on API Gateway Console

2. place it at swagger/swagger.json

3. Run Swagger UI using Docker

$ docker-compose up

5. Visit localhost:80

Usage

1. Verify your email on Amazon SES Console

If you are not yet request increasing SES limitation, you can only send email to verified email.

2. open /infra/config/dev.toml and replace values for your environment

3. copy dev.toml file under infra folder with name .toml

$ cd infra
$ cp config/dev.toml .toml

4. Generate csv file by following intructions of Amazon Personalize Samples

5. Upload csv file to S3 Bucket

Create Similarity Items campaign

1. Create Schema

$ http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/schema name=my-demo-schema schema={"type": "record", "name": "Interactions", "namespace": "com.amazonaws.personalize.schema", "fields": [{"name": "USER_ID", "type": "string"}, {"name": "ITEM_ID", "type": "string"}, {"name": "TIMESTAMP", "type": "long"}], "version": "1.0"}

2. Invoke api to create Personalize Similar-Items campaign

$ http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/similar-items name=my-similar-items-model schema="arn:aws:personalize:ap-northeast-2:776556808198:schema/my-demo-schema" bucket="s3://demo-similar-items-67914/DEMO-similar-items.csv"

Invoke API to get recommendations

1. Invoke api to get list of campaign_arn and copy your campaign arn

$ http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/campaigns

2. Invoke api to get recommendations realtime

$ http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/recommend/similar-items  campaign_arn=arn:aws:personalize:ap-northeast-2:929831892372:campaign/my-similar-items-model
 item_id=323

3. Invoke api to get recommendations in batch

$ http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/batch-inference name=my-batch-job solution_version_arn=arn:aws:personalize:ap-northeast-2:929831892372:solution/my-similar-items-model/84e322ff num_results=150 input_path="s3://demo-similar-items-67914/batch/input.json" output_path="s3://demo-similar-items-67914/batch/output/"

Put events

1. Get tracking id for dataset group

$ http get http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/tracking name==user-personalization-baseline

2. Put event via API Gateway

http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/events tracking_id=a6006e6f-8623-4684-bda4-33bec98aade9 session_id=user-personalization-session-1 event_type=click user_id=242 item_id=88 sent_at=1596258382

Cleanup Resources

1. Invoke api to remove dataset-group

$ http post https://xyxyxy.execute-api.ap-northeast-2.amazonaws.com/dev/cleanup name=my-similar-items-model

Cleanup

destroy provisioned cloud resources

$ cdk destroy "*"