Extension talk:RDFIO/Template matching for RDFIO
Use Cases
editHey Ali -
I agree with Sam - looks great. We'll want to have a set of use cases to develop around, so we should schedule "use case elicitation/development" in the first couple of weeks, and probably "use case refinement" in weeks 3 and 4, overlapping with the beginning of coding. I'm thinking the use cases will start with the most simple, and move up in complexity. I think that the most simple is: "pages already exist for each subject of a triple in the input; templates already exist for each class that those subjects are typed to; each page being edited only calls a single template."
Samuel and I are both involved in active development of SMW instances, so we'll be able to work with real data, and possibly use it to create unit tests. Also, our use cases will be motivated by the requirements that Sam and I have already gathered for our respective wikis. In my case, we are prototyping a botanical knowledge portal that will integrate trait data on plants with specimen data, genomic sequence data, and annotations made about all of the above via the W3C's Open Annotation ontology.
In terms of the benefits of this project, I think the biggest will be making SMW more interoperable with the rest of the semantic web. Related to this, I like that you mentioned Wikidata. Directly tying to Wikidata will probably be out of scope, but we should ensure that our approach is compatible with the Wikidata RDF model.
Looks great
editHello Ali, This looks great! Nice writeup!
The only little thing I wanted to ask more about is: "This requires the initial mapping of the most common ontologies (such as FOAF) to Semantic Mediawiki template types"
What is your thinking about how to do this mapping?
You might have some good ideas here already, but otherwise I just wanted to share my 5c on this: The method that has been used as much as possible throughout RDFIO, is to connect together wiki properties with their corresponding ontology URI:s, by the use of the Equivalent URI special property, and then do all the inferential logic and lookups based upon that. I imagine this should be possible to use also for Templates, by connecting a template's category to a corresponding OWL class, by the use of this same "Equivalent URI" property.