What is the best way for document level annotation?

[yagarea]

Hello,
We have set of documents we would like to annotate with document level annotations. We have created layers where layer represents group of document features. Some features are just picking one option from tag set and other are open and waiting for annotator to input a string.

Tag set settings example:

Layer settings example:

---

Problem

For annotator it looks like this:

• Its very unintuitive and slow adding each layer
• The required checkbox does not do anything because you can make annotations without those features (we are looking for a way to force annotators to fill every feature for every document)

Do you have any recommendations what to change in our project to make annotation workflow better?

[reckart]

I assume you find it unintuitive because in your setup, the annotators manually need to an instances of each layer.
Further, I assume that you are not actually interested in having multiple annotations per layer but rather only one instance of each layer.
If this is the case, enable the singleton option on the document metadata layers. This will pre-create an annotation of each layer and prevent the users from creating additional annotations. Then, the annotators only have to fill in the values.

You can enable the required flag for features of document-level annotations, but I would not recommend it at the moment. This is because curation of document-level annotations is not possible still and this can lead to a situation where the system internally prepares a document-level annotation for curation, but if there is no agreement, a required field may end up empty and the curator may not be able to fill in the required feature (in particular when using the split-view curation mode instead of the integrated curation mode). In the worst case, the curator can then not mark the document as "curation finished".

[yagarea]

Thank you very much for your answer. Singleton option fixed most issues with annotation workflow for us. We will probably do curation externally (export project -> check in another program) so we are not limited if the curation in inception is will not work as intended.

Thank you again for your support