Machine learning is available for auto classification as a cloud service referred to as Machine Learning Auto Classification (MLAC). This service provides assisted or automated classification of products into the primary product hierarchy and/or into classification hierarchies where the product can exist only once.
The MLAC service uses supplied hierarchies and products already classified to make suggestions, or predictions, for classifying products that are not yet classified. The topic Classification Predictions (here) shows the format of the prediction service requests and responses while the following topics describe how the prediction functionality can be integrated into / tested from STEP:
- Gateway IEP Configuration for MLAC (here)
- Business Action Configuration for MLAC (here)
- Testing MLAC Service Predictions (here)
- Integrating MLAC Predictions in a Workflow (here)
Before prediction requests can be made, the service algorithm must have been trained using customer specific data. For initial evaluation of the service, data for hierarchies and already classified products can be submitted manually to Stibo Systems. Manual data export is covered in the following topics:
It is, however, also possible to configure STEP to upload the data to the service directly. This functionality is covered in the topics:
Note: For information on standard automatic classification functionality without the cloud service, refer to the Automatic Classification documentation here.
The diagram below shows a high-level overview of the MLAC architecture and how MLAC can be configured into STEP.
