RE: Re 2: More role-related potential changes to Web Annotation Data Model

I do not have strong opinions about this, and I am going to say this poorly (imprecisely), but I think one concern stems from the fact that the oa:EmbeddedContent class actually can be used for more than just the range of oa:hasBody and oa:hasTarget (the definition of the class given in 3.2.4 of the data model probably needs to be improved), e.g., when you want to embed an oa:SvgSelector, as illustrated in Example 43 of the current Web Annotation Data Model, an instance of oa:EmbeddedContent can be used to embed the SVG of your selector as a literal string.  When used this way, the instance of oa:EmbeddedContent should not have a role property. But if you say in your RDFS/OWL ontology that oa:EmbeddedContent is the domain of oa:hasRole, there is not a good way to exclude the use of oa:hasRole when the instance of oa:EmbeddedContent is the SVG literal of an oa:SvgSelector.  For inferencing and other Semantic Web reasons, a subclass is needed for one or the other use.


I think this is an issue for inferencing and other Semantic Web use cases, but you probably have a better sense about this than me.


-Tim Cole



From: Ivan Herman [] 
Sent: Monday, September 28, 2015 5:30 AM
To: Tim Cole <>
Cc: W3C Public Annotation List <>
Subject: Re 2: More role-related potential changes to Web Annotation Data Model


Having looked at the text again, I question came to my mind. Let us suppose we go with Proposal 2. What that entails is to define a subclass of EmbeddedContent called EmbeddedTextualBody. However… the current OA draft specifies oa:EmbeddedContent in a section entitled "Embedded Textual Body"[1] and the class is used only to 'qualify' texts with media types or languages, etc. Why would we need a separate oa:EmbeddedTextualBody? Why isn't it simply enough to have and use one class (whether its name is EmbeddedContent or EmbeddedTextualBody is besides the point)?






[1] <> 




On 24 Sep 2015, at 23:18 , Timothy Cole < <> > wrote:


A new discussion document is up on GitHub: <> 


This a companion to the roles.html and AnnoLevelMotive.html documents in same folder.  Comments, corrections, etc. all welcome.  We followed a table of contents based on roles.html document, but if augmentation with additional use cases is needed, feel free (just don't delete or renumber any of the existing items in Table of Contents Section 3). 


Issue addressed by this new document: 

You will recall that earlier this month the WG reached a consensus to use a new property, oa:hasRole, for expressing roles of oa:SpecificResources serving as Annotation Body or Target.  But a couple of issues summarized in Section 3.2 of the roles.html [1] document were not formally resolved, specifically, the 'further considerations' discussed in:


*         3.2.1 Require the use of SpecificResource for Bodies

*         3.2.2 Require the use of SpecificResource for Targets

*         3.2.3 Allow hasRole on new EmbeddedTextualBody class


This new document captures some of the discussions we had around these further considerations and illustrates (through 30 examples, each in JSON-LD and Turtle) some of the implications of our options with regard to these 'further considerations' and with regard to how role might interact with multiplicity classes (probably an edge case).  


At this point, Paolo and Rob are already working on the next update of the Web Annotation Data Model, so it'll be up to them and Frederick whether to revisit these further considerations now or wait until after the next iteration of the Data Model is ready, or at least until TPAC. It may be that the discussions the WG has already had and the process of updating the model will clarify things to such a degree that we don't need to revisit these issues during one of our upcoming calls/meetings – in which case the new page will just help complete the documentation of the hasRole discussion we had over the summer. But if more guidance from the WG on these further considerations is needed at this time, the page is available to facilitate that discussion. 


[1] <>  



Tim Cole


Ivan Herman, W3C 
Digital Publishing Lead
Home: <> 
mobile: +31-641044153



Received on Monday, 28 September 2015 16:22:13 UTC