Feedback on OAC Data Model

[Robert Sanderson]

Dear all,

We have been discussing the feedback provided on the Alpha 2 data model, and have come up with some changes that we hope address the majority of the concerns.

In particular:

  * Annotations are now modeled as information resources, specifically as RDF documents. They are no longer non-information resources that need to be described by something else.

  * As a consequence, the model is no longer based on ORE.  Annotations are not Aggregations, and the ORE Proxy construction is no longer used.

  * Segments of resources are modeled as a resource with a unique identifier, and hence can be used as the subject or object of an RDF triple.


Please see:
  http://annotation.lanl.gov/alpha3/alpha3.pdf 
for a slide deck explaining the revised data model.

We would be very grateful for any feedback, however there is one issue in particular which we would like to hear your opinion on:

In the slides above, there is only one link between the Annotation and the Target, whether or not that is a full resource or a segment of a resource.  It has been suggested that in order to make client development easier, there should be a link directly from the Annotation to the full resource as well as to the actual target.  The counter argument is that if the client can parse RDF and follow at least one link from the Annotation, then it can very easily follow the next link to find the full resource, without cluttering the model with redundant links.

The changes this would cause for the model are presented in:
  http://annotation.lanl.gov/alpha3/alpha3_options.pdf

Sharing any experience regarding this sort of problem, both at design and implementation levels, would be greatly appreciated.

Many thanks,

Rob Sanderson for the OAC team

[Daniel, Ronald (ELS-SDG)]

Changes sound very encouraging Robert.

 

Re. your question on having a link to the full resource when the Target is a fragment:

Our experience has been to go ahead and provide the redundant information. Reason for that is that we have to swap data with a lot of people who are not RDF experts. One example of such would be when there is an error as people try to move content over various links. We may end up on the phone with someone, and the tools they have available are simple text editors. It is easier to do that debugging when there is redundant, and human-readable, information available.

 

Ron