The presence of the human validation element in the typical workflow means that a feedback loop can be created between the AI model predictions and the corrected human validations. With this feedback loop, a self-learning mechanism can be set up which means an Extractor continues to learn and improve over time based the review conducted by users on the platform.

Enabling 'self-learning' capability

Any tailored or custom model can have the self-learning capability enabled. 'Base' models cannot be self-learning as they are not customer-specific so controls are in place to ensure model regression does not occur.

Self-learning is toggled at a Collection level. When toggled 'on', all documents confirmed by a user from this Collection will be sent to the Extractor attached to this Collection for training. This gives users ultimate control over what documents get fed into the training queue for a specific Extractor and means that test Collections can be used to test model performance or other features without impacting on the performance and accuracy of a given Extractor.