In each of these cases, the annotation (the note, the keywords or the grant) are used to supplement the information of the underlying dataset. We consider this a “data-focused” annotation system, so the data record is most commonly the object being annotated. It is possible that an annotation can itself be annotated, or the body of an annotation.
An annotation assumes that each body applies to each target equally. This may not be the case, for example, we may want to generate an annotation for a dataset that both links it to a record from another database, and explains why the records are being linked. In this case, neither of the body elements can be considered independent of one another with respect to the target since the text must reference the URI and the URI relationship is not clear until addressed by the text annotation.
-The OpenAnnotation standards provide the concept of a Composite. In this case, the two body elements can be bound together, indicating that they must be considered in connection to one another, and then linked to the annotation.
-body:(composite:((body1)(body2)))-annotation-target:()
The OpenAnnotation standards provide the concept of a Composite. In this case, the two target elements are bound together, indicating that they must be considered in connection to one another, and then linked to the annotation body that expresses the relationship.
+body:()-annotation-target:(composite:((body1)(body2)))
The annotation requires some secondary information if we are going to manage annotations across the data lifecycle. These include the creator, the created date/time stamp, the generator, the agent responsible for generating the reference, the date in which the serialization was generated, and then, potentially, modified.