Draft rNews 1.0 OWL Ontology: Feedback Requested
We’ve created an IPTC rNews 1.0 ontology file in OWL. It is available at
http://dev.iptc.org/files/rNews/rnews_1.0_draft1.owl
This is just a draft, but we’d like to get feedback from the experts on any improvements we can make, whether it is in syntax, structure, documentation or any other aspect.
So, let us know what you think!
Regards,
Stuart
Hi
I am Bahareh Heravi working at DERI Galway on the Social Semantic Journalism project. The project mainly looks at the ways in which Semantic Web technologies could assist journalists in the news production life cycle, mainly in relation to user generated content available/received on the Social Web and other publicly available data sources, e.g. archives and linked open data.
I was just wondering about the reasons behind choosing an OWL representation for rNews, but keeping it as simple as an RDF(S) formalism? Why not RDF in this case? I have to say I am somehow an OWL fan because it is more expressive, but believe perhaps RDF(S) should be used when OWL expressivity is not needed/used?
My other question would be on whether rNews plans to take into consideration the news posted by citizen journalists on the social media platforms and further if it plans on using any of the relevant, widely used, ontologies such as FOAF or SIOC?
Cheers
Bahareh
Hi Bahareh,
I produced this first draft and just like you, I like OWL’s expressiveness. Don’t really see the need to pick a different representation just because I’m not using all the features right at this moment, but that might also be a conversation for a different forum.
My other reason for using OWL was that while it’s currently very simple (and probably still full of questionable choices I made because I’m not an OWL expert), I wanted to lay a foundation to expand it if the community so chose.
We have addressed your question regarding other ontologies at length in our talks and communications during the rNews feedback phase leading up to version 1.0. The gist of it is: no, rNews will not use those other ontologies to keep implementation simple.
But this OWL file could – and should! – absolutely link rNews to those other ontologies, where appropriate. That’s precisely the reason why we want everyone’s feedback.
Would you like to take a stab at mapping rNews properties to FOAF or SIOC? Please let us know.
Appreciate your interest,
Andreas
Thanks Andreas,
I am more than happy to work on mapping rNews to SIOC, since SIOC is developed in my unit at DERI. I am also happy to assist in the mapping of FOAF.
I would also be working on an ontology for the Social Semantic Web project and would like to make sure the relevant ontologies and vocabularies, including rNews and Schema.org, are carefully taken into consideration, and I guess the mapping would help in that respect too.
Bahareh
Here is the second draft of the rNews 1.0 OWL ontology. It’s still a work in progress.
Turtle: http://dev.iptc.org/files/rNews/rnews_1.0_draft2.ttl
RDF/XML: http://dev.iptc.org/files/rNews/rnews_1.0_draft2_rdfxml.owl
I addressed the following issues:
– cleaned up typos in some URIs (rnews vs. rNews)
– aimed to properly declare that ImageObject != AudioObject != VideoObject (although I can’t seem to declare a domain or range as a disjointUnionOf, because Protégé and other tools seem to have issues with that… any thoughts are welcome, I’m sure it’s a rookie mistake I’m making)
– add isDefinedBy for most of the classes (one is missing an appropriate resource to point to at this time)
Please have a look at the above linked files and suggest fixes to errors you encounter. It has been suggested we validate the file against OOPS [http://www.oeg-upm.net/oops] which is a great idea but the service is currently down and I still wanted to get the second draft out. So.
Lastly, we’d like to get the rNews ontology scrubbed of errors before we begin adding mappings to other well-established ontologies. (FOAF and SIOC have been suggested thus far.)
General rule: if you suggest a mapping, please be prepared to help. 😉
Feel free to email me directly or send me a tweet
a few encoding issues:
rdfs:comment “A secondary headline or title of a NewsItem.”@en ;
rdfs:comment “A count of a specific user interactions with this item—for example, 20 UserLikes, 5 UserComments, or 300 UserDownloads. It is suggested (but not required) that users draw the interaction type from the controlled vocabulary provided by schema.org at: http://schema.org/UserInteraction“@en ;
articleBody, commentText rdfs:range xsd:string .
no HTML allowed via rdf:XMLLiteral?
The range of discussionUrl is xsd:anyURI or xsd:string but replyToUrl is xsd:anyURI only. would make sense to maybe set the range of discussionUrl to xsd:anyURI only.
rnews:url rdfs:comment “A reference to a web resource.”@en ;
rdfs:domain ;
what’s the purpose of this url property associated to a PostalAddress? the rdfs:comment could be more explicit.
have you considered xsd:duration as range of duration?
rnews:Article rdfs:isDefinedBy .
My understanding is that isDefinedBy is meant to point to the RDF vovcabulary defining this class, in other words the rnews namespace. for an HTML documentation page, maybe rdfs:seeAlso? see http://answers.semanticweb.com/questions/2149/what-is-the-intuitive-difference-between-rdfsseealso-and-rdfsisdefinedby
Here is the third draft of the rNews 1.0 OWL ontology:
Turtle: http://dev.iptc.org/files/rNews/rnews_1.0_draft3.ttl
RDF/XML: http://dev.iptc.org/files/rNews/rnews_1.0_draft3_rdfxml.owl
I addressed the following issues:
– cleaned up encoding issues mentioned by Stéphane
– fixed isDefinedBy declarations
– enhanced some comments to provide additional clarity
– added strong AND weak datatypes, to declare the most precise option for those who can/want to use them but also account for the realities of the web
I also validated the file against OOPS [http://www.oeg-upm.net/oops] which returned mostly missing rdfs:label and some missing rdfs:comment — both of which we’ll fix soon. It also complained about some missing inverse relationships, which I’d like to hear your feedback on: We did not explicitly add relationships like e.g. “wasCreatedBy” (as an inverse to “creator”) or “wasEditedBy” (as an inverse to “editor”) because we felt that could be inferred and it would otherwise bloat the data model unnecessarily. Your thoughts on that, please?
Please have another look at the above linked files and suggest fixes to errors you encounter.
At this time, we’re planning to add mappings to these other ontologies:
– schema.org (we’ll provide)
– FOAF (Bahareh Heravi volunteered (thank you!))
– SIOC (Bahareh Heravi volunteered (thank you!))
Any others you’d like to see? Suggest? Volunteer for?
Feel free to email me directly or send me a tweet.