Training

Training, or Model Training, is the process by which a user teaches Communications Mining what labels and entities apply to messages, so that it can apply this understand at scale across the entire dataset.

A user trains by manually reviewing messages and applying all the labels and entities that are relevant.

Each time a user spends some time reviewing messages, it triggers a re-training event where the platform uses the newly reviewed messages to improve its understanding of the label concepts and entities.

After each re-training event, the model re-reviews all of the unreviewed messages in the dataset and updates the predicted labels and entities and their associated confidence scores (and sentiment for labels if label sentiment is enabled).