To build a custom classifier, you need to upload a set of unlabeled documents through the Data sources tab. Alternatively, you can use one of the publicly available datasets available through the Samples library tab.
Move into the Samples library tab where you can select a sample dataset from publicly available data. Select Customer Support Tickets and press "Continue".
Preview a sample of the data. Press "Continue".
Provide a name and a description for your customer support ticket classifier and click "Build Model". Now you will start teaching the machine about the semantics of the categories on your dataset.