Two new features have been added to enable users to create their own custom models within the service:

• custom key value extraction
• custom document classification.

With these additions, customers can easily tailor key value extraction and classification models to their industry or use-case specific documents. These features are available in all OCI commercial regions and can be accessed through OCI software developer kits (SDKs) and REST APIs.

Key-value models extract specific, predetermined values from documents. To create a custom key-value model, users supply their own training documents and then use OCI Document Understanding to specify which text correspond to each key label. The labeled data set is then provided to Document Understanding , that then performs the training and produces the customer model that can them be consumed through the API or SDK.

Custom classification models are similar except that each document is labeled with a single classification category representing the document type. The trained custom classification model can then recognize the types of documents that it was trained for.

For more information on custom models, see the Document Understanding documentation,