Medical annotations capture a structured representation of knowledge stored in unstructured text. The task of mapping text to structured knowledge is done with the end goal of feeding the annotations into a machine learning algorithm that learns how to automatically extract the medical knowledge contained in the text. The guidelines defined below establish an annotation standard to be followed by human annotators.
The Guidelines cover the annotation methodology, including recommendations for training and organizing the annotations team, as well as specific guidelines and examples for annotating medical entities, modifiers, entity temporal and certainty assessments, and entity relations.
By sharing this document with the broader community we encourage researchers to follow the standardized approach for data annotation, hence producing high-quality medical text annotations.
2. Annotation Guidelines (introduction)
- Defining Entity Temporal Assessment
- Defining Entity Certainty Assessment
- Rules for Annotating Temporal Assessment, Certainty Assessment, Subject