EDUCAÇÃO E TECNOLOGIA

Announcement: Invoice Object Recommendation becomes part of Data Attribute Recommendation – Technical Details (Part 2)


Introduction 

As announced in our previous blog post, we decided to merge the Invoice Object Recommendation service into the Data Attribute Recommendation service as a business blueprint. Let’s have a deep dive into the technical changes starting from December 2021 after the merger of Invoice Object Recommendation and Data Attribute Recommendation.
 

Why is it happening? Benefits from the Merger 

With the announced merger of the Invoice Object Recommendation service into Data Attribute Recommendation, the functionality of Invoice Object Recommendation continues as a business blueprint in Data Attribute Recommendation, that would serve your business case as it used to be. Customers can now choose the Invoice Object Recommendation business blueprint model for training the data they have already uploaded. 

The Invoice Object Recommendation business blueprint – as of December 2021 – leverages the Data Attribute Recommendation capabilities which sequentially will benefit SAP customers like one single solution for classification tasks or full model lifecycle. If you would like to read more on the benefits of this merge, please read the first part of this blog. 

What is Changing 

Here is a deeper look at what customers can expect when using the Invoice Object Recommendation business blueprint. The below table compares the main aspects in the context of handling data: 

    Invoice Object Recommendation Service  Invoice Object Recommendation Business Blueprint 
Functionality  Features 

Classify G/L Accounts (HKONT), Cost Objects (KOSTL), and CO-PA Dimensions (COPA) 

 

Classify G/L Accounts (HKONT), Cost Objects (KOSTL), and CO-PA Dimensions (COPA), additionally adding more features and labels possible 
Applications & Endpoints  Single application with the same base URL   Multiple applications (three) with multiple base URLs (data management, model management, inference) 
Data Upload  Data model management  Input data is already defined by the service  Create dataset and define dataset schema to fit requirements of the business blueprints, in addition to adding more labels 
Type  Upload data set: .CSV  Dataset schema: .JSON
upload data set: .CSV 
Pre-processing  Handled by SAP’s side (data filtering, modelling, processing, feeding in the training pipeline)  Handled by SAP’s side (data filtering, modelling, processing, feeding in the training pipeline) 
Conditions, batch upload  

Only limited to 20MB per upload (batch upload) 

File support: .CSV  

Up to 5GB (no batch upload) 

File support: .CSV and zipped .CSV 

Dataset lock/Dataset deletion  The data set is locked after all batch uploads. Not able to delete the datasets nor upload batches till lock is removed/training is completed  No data lock is required. Not able to delete the dataset unless you delete the model and the training job 
Training  Dataset selection  N/A  Enter the Dataset’s ID for the  model to be trained   
Template selection  N/A  Choose Invoice Object Recommendation Business Blueprint 
Trigger training   N/A. Trigger the training by providing the job ID  After choosing the template and the dataset  Dataset ID and providing a model name,you can trigger the training 
Metrics  Training accuracy and test accuracy  Test accuracy, F1 score, precision and recall 
Inference  Format  Input and Output file format: .CSV  Input and Output file format: .JSON 
Parameters  N/A  Can be defined 

 

It is also good to visualize the differences in the process flow through the below diagram: 

Please use our dedicated Q&A section in SAP Answers to ask questions about both Data Attribute Recommendation and Inovice Object Recommendation going forward 

Visit our SAP Community page

Read our product documentation