After you create enriched datasets or machine learning results in a partner tool, you can publish them as data products to SAP Datasphere.
This process allows you to integrate these new insights into your analytical models, intelligent applications, and AI scenarios.
Because the steps for generating a data product depend on the specific tool you use, refer to your tool's documentation for detailed instructions.
To demonstrate the workflow, the following section uses SAP Databricks as an example.
A primary use case for sharing SAP Databricks results back to SAP Datasphere is to enrich standard SAP-delivered data products with machine learning data. To do this, you first generate a custom data product from your SAP Databricks machine learning results. Then, you build a data model in SAP Datasphere that combines the standard SAP data product with your custom data product.
To share machine learning results from SAP Databricks with SAP Datasphere, you must first store the results in a delta table. Once the delta table is available in your SAP Databricks catalog, you can create a delta share and subsequently a custom data product.
Create the Delta Share
To create a new delta share in SAP Databricks, open the Catalog Explorer and select the Manage icon to access the Delta Sharing management page.
In the Delta Sharing section, navigate to the Shared by me tab and select the Share data button.
In the Create share step, provide a name and a description for the new share.
In the Add data assets step, select the delta tables you intend to share.
You can skip the Add notebooks step if you are not sharing notebooks in this scenario.
In the Add recipients step, select the sap-business-data-cloud target system.
The new share is now available within your SAP Databricks workspace.
Create the Data Product
To use the delta share as a data product in SAP Datasphere, you must create two specific files using Python code before publishing.
An Open Resource Discovery (ORD) file is required to describe the resources and capabilities of your data product. The ORD protocol is an open standard originally developed by SAP. It allows compliant data providers to expose data products and enables consumers to discover them.
A Core Schema Notation (CSN) file is required to define the structure of the data product, including entities, properties, and relationships. The CSN file acts as the official contract for the data, providing a consistent definition for both developers and automated systems.
After creating these files, you can publish your data product.
Access the New Data Product in SAP Datasphere
Once the data product is shared from SAP Databricks, you can access it directly in SAP Datasphere.
The data product will be listed in the SAP Datasphere catalog.
You can then install it in an SAP Datasphere space and consume it like any other data product.
Summary
- Create a delta share in the external tool by selecting delta tables and designating SAP Datasphere as the recipient.
- Develop ORD and CSN files to describe and structure the data product for SAP Datasphere.
- Publish the data product to SAP Datasphere to enhance standard data products with newly generated insights.
- Access and install the data product in SAP Datasphere for consumption.