Authentication

Client systems gain access to SAP Service Ticket Intelligence API through an OAuth 2.0 access token request.

post

Obtain the bearer token for use in subsequent training and prediction calls. The token has a limited validity and needs to be re-requested upon expiry.

Model Training

Model Training API are generic across different Machine Learning services such as Text Classification and Solution Recommender, and refer to common Machine Learning model attributes such as Training File, Model Status and Accuracy.

post get delete

post

Upload a training file to create new ML models. The ML configuration settings contain a set of four items: scenario, field mapping, topic tabulating, and training settings. Uploading a training file results in a model ID to be used to create its corresponding machine learning model.

get

Retrieve the list of models in the system.

delete

Remove a machine learning model or an uploaded training file to save storage space. Models that are ACTIVE or IN_TRAINING cannot be deleted.

post

Upload an incremental training file to create new ML models from existing ML models. The incremental ML configuration settings include a reference model ID and training settings.

get

Retrieve the training results (model details) using the model identifier obtained in List Models or Upload Training File endpoint.

post

Initiate the training process for a training file which has been uploaded previously.

get

Retrieve the validation results for a given model identifier.

put

Activate a trained model so that it will be used subsequently by inference calls to Text Classification or Solution Recommender. This implicitly deactivates the current active model (if any) for the same language and business object. Only models whose status is READY can be activated.

put

Deactivate a trained model, without activating another model. Upon successful execution of this request, the deactivated model returns to model status READY.

put

Disable a particular category class, so the category class will not be eligible for Text Classification inference.

put

Enable a particular category class that has been previously disabled, so it will be again be eligible for Text Classification inference.

Text Classification

Text Classification API categorizes unstructured customer service messages eg. email or social media, into user-defined categories such as question, complaint or comment, as determined from the historical service tickets. The classifcation results may be used to streamline the processing of the customer service ticket, for example automate ticket assignment.

post

Classify incoming customer service messages.

Solution Recommender

Solution Recommender API makes recommendations for unstructured customer service messages based on historical service ticket content and the corresponding knowledge base items that were used to solve it. Two types of recommendations are possible: “question_recommendation” and “answer_recommendation”, depending on the scenario defined during Machine Learning model creation.

post

Recommend possible solutions for incoming customer service messages.