Human Annotation in KAPTO

In KAPTO, we understand the value of human touch in refining the work of artificial intelligence. While our AI-powered extraction models are proficient at identifying and extracting entities from a document based on a predefined schema, we also offer the ability for these AI-extracted entities to be reviewed and amended by a human operator. This process is commonly known as human annotation.

Human annotation in KAPTO serves to further enhance the precision of the extracted information. Each entity identified by the AI can be validated and, if necessary, corrected by a human. This creates a system of double-checking where the speed and scalability of AI are coupled with the understanding and discernment of a human reviewer.

The human is asked to annotate AI values only when explicitly requested by KAPTO. This ensures that the review process is both efficient and targeted, focusing human efforts only where they are most needed.

For every identified entity, there can be two types of values: the AI-extracted value and the human-annotated value. The AI-extracted value is the original entity value that the AI model identifies and extracts from the document. The human-annotated value, on the other hand, is the reviewed and possibly amended entity value provided by the human operator during the annotation process.

By balancing the power of AI with the accuracy of human judgement, KAPTO offers a robust mechanism to ensure the highest quality and accuracy in entity extraction. It combines the strengths of both AI and human intelligence, thereby optimizing the entity extraction process.