Display annotations by server-specified type

[benwbrum]

Annotations provided by other systems may fall into categories that do not really fit into motivation or other standard parts of the spec. For example, a manuscript might have annotations representing a diplomatic transcription, an interpretive transcription, a translation, and perhaps an index or commentary.

If Mirador could display annotations of a single one of these categories alongside an image, we could use it to display a bilingual edition as a triptych, with a panel for the image, a panel for the transcription, and a panel for the translation, or as a toggle between transcription and translation within a single panel. (See The Dream of the Rood in EVT for an example of a toggle UI in a non-Mirador viewer.)

A simple example of a (v2.1) manifest with a transcript and a translation produced by FromThePage is https://fromthepage.com/iiif/34796/manifest, in which each canvas contains separate annotation lists for transcriptions and translations. (See the corresponding document in FromThePage.)

A more scholarly example is the Deiphira Project, which is transcribing and translating Deiphira on FromThePage with a corresponding manifest at https://fromthepage.com/iiif/33728/manifest . In this case, using a Mirador workspace as a publication platform (perhaps using a content state that opened panels for transcription and translation?) would be very appealing.

Modeling options

If Mirador allows this kind of filtering by category, how should annotation providers model their data?

Annotations of Diplomatic Transcription, Interpretive Text, and Translation categories might have a motivation of commenting, supplementing, or both, but there is no way to use motivation to differentiate them from each other. (Even if there were, it probably shouldn't be the job of the WA spec to enumerate every conceivable scholarly output.)

One possible option would be to use purpose to differentiate annotations (or annotation bodies) from each other in this way. The down side of purpose is that it seems hard to attach human-readable metadata to each category of annotations, which is particularly challenging for internationalized labels.

Perhaps a better way would be to use OA layers (now AnnotationCollections) to aggregate annotations of the same category. A canvas could have an AnnotationCollection for "Diplomatic Transcription", "Normalized Text", or "Translation", each containing only the annotations of that category. This would put much of the burden of presentation on the annotation provider (who has the domain knowledge aggregate and label appropriately), so that Mirador would only need to present a AnnotationCollections as options and then display annotations from a collection.

See related issue #2946