Train Data Attribute Recommendation Service Machine Learning Model by Connecting to SFTP using SAP Cloud Integration (Part 1)
Introduction to Data Attribute Recommendation Service
There are many business scenarios in the enterprise, where classifying various kinds of data, either master data or transactional data is an important task. For instance, during master data creation of a product, it is very important to classify a product into the right category and product hierarchy. In general, this is a manual process and only someone with the knowledge of the products can do this accurately.
To help automate this manual process, Data Attribute Recommendation service, which is part of the SAP AI Business Services portfolio, can learn from historical data of products and their hierarchies to automatically suggest the relevant categories with confidence level. This saves a lot of manual effort and helps automate the master data creation process. This same concept can be applied to the completion of transactional data like sales orders or any other orders with missing information to complete the same. For more info, read the blog post Machine Learning inside Data Attribute Recommendation
The first step towards using the Data Attribute Recommendation service for your use case is to train the machine learning model. Let’s look at those details.
Machine Learning Model Training in Data Attribute Recommendation
In order to train a Machine Learning model using Data Attribute Recommendation, you would first need to create a dataset, dataset schema, upload training data in CSV file format and then create a training job to train the model. All of these steps can be done using REST APIs. You can use tools like Postman as described in tutorial to do the training. One of the better approaches for enterprise architecture is to use a solution like SAP Cloud Integration to carry out all these tasks, rather than manually doing them using Postman.
Using the recently released SAP Cloud Integration Package for Data Attribute Recommendation, you can create datasets, upload data and train machine learning models, including deploying the same.
In the next sections we will look at how to use the SAP Cloud Integration package for Data Attribute Recommendation, including configuration, and executing the training and deployment.
Setup of Data Attribute Recommendation service instance
As a first step to use Data Attribute Recommendation, you need to setup the service instance and have the service key information. You can use the tutorial to setup Data Attribute Recommendation service instance using the interactive booster.
SFTP Server Credentials
As the CSV files that are used for Data Attribute Recommendation model training would be stored in an SFTP server, you would need access to an SFTP Server and the SFTP Server user credentials before you proceed further.
After you have the service instance setup and have the service key information, including the SFTP user details handy, follow the steps in next section to setup the SAP Cloud Integration Flow.
Configuration of the SAP Cloud Integration Flow
The integration flows which integrates Data Attribute Recommendation are available on SAP API Business Hub.
In order to activate and use the integration flows you need to have SAP Cloud Integration enabled. In case your SAP Cloud Integration is not setup you can follow this tutorial.
If this prerequisite is met, you can follow the steps below to configure the SAP Cloud Integration for Data Attribute Recommendation.
User Credentials Setup
You need to access the “Manage Security Material” section in order to create the new user credentials for the Data Attribute Recommendation integration:
1) Go to “Monitor” >> “Security Material” under “Manager Security” section of SAP Cloud Integration as shown in below screenshot.
2) Click “Create” and select “User Credentials”.
3) Add User Credentials for the Data with the following settings and click “Deploy”:
- Input Name as <any name for Data Attribute Recommendation user>
- Input Description as <any description for Data Attribute Recommendation service key>
- Select Type as User Credentials.
- Input User with <clientid> from the service key JSON file.
- Input Password with <clientsecret> from the service key JSON file.
4) Add User Credentials for the SFTP Server with the following settings and click ”Deploy”:
- Input Name as <any name for Data Attribute Recommendation SFTP User>.
- Input Description as <any description for Data Attribute Recommendation SFTP User>.
- Select Type as User Credentials.
- Input User with <username> for the SFTP Server.
- Input Password with <password> for the SFTP user.
The next step is to discover and copy the integration flow artifacts
Discover and Copy the Integration Package
In this step we discover and copy the integration package for integration package for Data Attribute Recommendation Service into your SAP Cloud Integration tenant.
1) Navigate to the “Discover” section of SAP Cloud Integration and Search for “SAP AI Business Data Attribute Recommendation” and copy the package with name “SAP AI Business Service Data Attribute Recommendation”.
2) Once the package is copied, go to the “Design” section and select “SAP AI Business Service Data Attribute Recommendation”.
3) In the package, “SAP AI Business Service Data Attribute Recommendation”, you should be able to see 3 integration flow artifacts as shown below:
In the next part of the blog, we will review how to configure each of the integration flows and train the machine learning model using the SAP AI Business Service Data Attribute Recommendation.
In the meantime, you may read how to integrate your enterprise email inbox with the Document Information Extraction service via SAP Cloud Integration
For more information on SAP AI Business Services:
Explore: SAP Community Page
Dive deeper: Open SAP Course
Get an overview: Blogpost part I | Blogpost part II
Document Classification Questions | Document Information Extraction Questions
Business Entity Recognition Questions | Service Ticket Intelligence Questions
Data Attribute Recommendation Questions | Invoice Object Recommendation Questions
Credit to Jana Wuerth who helped me review this blog post