CLAIRE Server Installation for ML Model Driven Use Cases
Hardware Requirements
At least 4 CPU cores
At least 10 GB of RAM, free to use by the CLAIRE recommendation service
Windows Installation
To install the CLAIRE recommendation service, an installation script is provided. It's located in the CLAIRE accelerator package under server.ai/installation. It will automatically install Python and all the required machine learning libraries in the needed versions for the CLAIRE recommendation service.
All you need to do for this is to follow below steps:
Open File Explorer
Locate the installation folder with the script
Run the script installation-script-windows.bat by right clicking the batch file and selecting the Run as administrator option (some of the commands require administrator privileges)
After you complete the steps, the batch will run each command in sequence displaying the results in the command window.
Folder creation
In addition, you have to create 4 folders
c:/informatica/ai/data/product_classification here is where the data used for the product classification training process will be saved
c:/informatica/ai/model/product_classification here is where the product classification trained models used for the classification process will be stored
c:/informatica/ai/data/attribute_extraction here is where the data used for the brand extraction training process will be saved
c:/informatica/ai/model/attribute_extraction here is where the brand extraction trained models used for the brand extraction process will be stored
The folders can be placed in a different location, and you can configure the path within the server.ai configuration file (config.ini). See chapters Configuration and Operation of CLAIRE Product Classification under the section Configuration of CLAIRE product classification service and Configuration and Operation of CLAIRE Brand Extraction under the section Configuration of CLAIRE brand extraction service for details.
CLAIRE server start up
Finally, you can start the CLAIRE server. A start up script Start-CLAIRE.bat is located in the CLAIRE accelerator package under server.ai. Navigate there via your Command Prompt and start the script by typing in Start-CLAIRE.bat.
In a few seconds, you should see the server up and running as shown in the screenshot below.
Linux Installation
To install the CLAIRE recommendation service on a Linux machine, an installation script installation-script-linux.sh is provided. It's located in the CLAIRE accelerator package under server.ai/installation. It will automatically install Python and all the required machine learning libraries in the needed versions for the CLAIRE recommendation service.
All you need to do for this is to follow below steps:
Navigate via your shell to the server.ai/installation folder
Run chmod +x installation-script-linux.sh to make the installation script executable
Run ./installation-script-linux.sh
After you complete the steps, the script will run each command in sequence displaying the results in the shell window.
Folder creation
You have to create 4 folders
/tmp/informatica/ai/data/product_classification here is where the data used for the product classification training process will be saved
/tmp/informatica/ai/model/product_classification here is where the product classification trained models used for the classification process will be stored
/tmp/informatica/ai/data/attribute_extraction here is where the data used for the brand extraction training process will be saved
/tmp/informatica/ai/model/attribute_extraction here is where the brand extraction trained models used for the brand extraction process will be stored
The folders can be placed in a different location and you can configure the path within the server.ai configuration file (config.ini). See chapters Configuration and Operation of CLAIRE Product Classification under the section Configuration of CLAIRE product classification service and Configuration and Operation of CLAIRE Brand Extraction under the section Configuration of CLAIRE brand extraction service for details
CLAIRE server start up
Finally, you can start the CLAIRE server. A start-up script Start-CLAIRE-linux.sh is located in the CLAIRE accelerator package under server.ai. Navigate there via your shell and start the script.
In a few seconds, you should see the server up and running as shown in the screenshot below.
Claire Server Health Check
To check whether the python server is up and running and the CLAIRE application can respond to HTTP requests, we provide a REST API endpoint that you can test against:
|
URL pattern |
/health |
|
Method |
GET |
|
Success response |
|
Congrats, this should have set you up properly to work with the Informatica MDM - Product 360 CLAIRE recommendation service. Enjoy!