Recommending Suppliers and Questions

To use these intelligent recommendation features, you first need to train the machine learning models. To perform these duties, you must be a member of both the SV Enrichment Manager group and the Customer Administrator group.

SAP Ariba Sourcing can recommend suppliers to invite to a guided sourcing event based on similar past events.

Data used to generate the recommendation includes:

• Item name
• Event name
• Department
• Commodity
• Region
• MaterialNumber
• MaterialCode
• Supplier history in past events, including invitations and contracts

For commodity, region, department, MaterialNumber, and contracts, if the data used to train the machine learning model mostly includes values for these fields, they are used in the recommendation. If these fields are mostly empty in the training data, they aren't used in the recommendation.

In addition, if you have retrained the machine learning model, data used to make the recommendation also includes past experience with the model. For example, if the model recommends ten suppliers and the user chooses only six of them, then the next time the machine learning model is retrained, this information is used to refine the model.

The following procedure outlines how administrators can train the MLMs:

1. Manually Training the Supplier Recommendation MLM - Generate a model file to compile transactional data that helps the system learn potential field names, values, data types, etc.
   1. Choose Manage→Administration→Enrichment Manager→Model Wizard.
   2. In the Model Wizard page, select SupplierRecommendation as the enrichment type in the first step.
   3. Skip to the third step, Train and perform the following tasks:
      1. Specify the Last Modified Date range (recommended: one year to date).
      2. Select the Generate Training File button and confirm the generation of the file.
      3. Choose Refresh to view the newly-created file in the list of files on the page.

         A file that contains event-related transactional data for the specified time period is created.

   4. From the Actions drop-down for the newly-created file, select Post to training and confirm.

      The model training results appear in the Monitor section.

   5. In the Monitor section, choose Report to download the classification report for a completed job.
      Note

      The Report button is disabled for the following enrichment types: Supplier Recommendation, Content Recommendation, and Sourcing Unstructured.
   6. Select the Manage section to view, upload, or delete trained models.
2. Configuring Automatic Training of the Supplier Recommendation MLM - After you generate a model file manually, set up automatic update of the model file at preconfigured intervals. Automatic update of the file enables the system to update the mappings based on the most recent transactions.
   1. Choose Manage→Administration→Enrichment Manager→Model Wizard.
   2. In the Model Wizard page, select SupplierRecommendation as the enrichment type.

      SAP Ariba recommends that you set up monthly automatic retraining of the machine learning model for intelligent supplier recommendations.

   3. Select Configure to open the Model Generation Configuration page.
   4. Select Advanced Settings.
   5. Enable the Auto Train options and select Update.

      The Configuration settings updated successfully message appears.

To train the MLM for Content Recommendations, perform the same steps as before, but select ContentRecommendation for the Enrichment Selection.

• The MLMs for recommending suppliers and content is done within the Manage→Administration→Enrichment Manager→Model Wizard.
• The Enrichment selection for suppliers is SupplierRecommendation and for recommended content it is ContentRecommendation.