What Sucks About PlanetRDF.org
First off, let me mention that I think the idea of PlanetRDF.com is great, and I find it a very valuable resource.
My problem with planetrdf.com is that it syndicates all posts from blogs it has chosen, instead of those posts that deal only with RDF and the semweb. For instance, planetrdf.com might show posts on RDF and someone's new car. While I might be interested in that new car (especially if it's a Mini), I would like planetrdf.com to filter out items that do not have anything to do with the semweb or RDF. For a site that is all about RDF, I think we have all the tools necessary.
It should be straightforward for planetrdf.com to say, "This item (blog post) has something to do with one of the categories I syndicate. Let's show it.". Also, it should be just as easy to say "This item is from a category that I do not syndicate, so I'll silently drop it".
In RSS 1.0 (RDF) there should be a triple to indicate an item's category. From the implementations I've seen, the Object in this triple is a Literal (string). For instance, here is a triple you might get from a RSS 1.0 feed:
The most simple thing to do here is to create a list of approved Literal values of categories. The planetrdf.com aggregator would simply check the string values of the dc:subject.
Now, of course, we are using Semantic Web technologies, so we can do a bit better. The property dc:subject doesn't specify a rdfs:range. I'm not sure exactly what this means, but I'm hoping it means we could place a Resource as the Object of that triple.
For instance, the above triple could change to:
Then, using OWL, we can provide mappings between Resources that planetrdf.com is interested in (such as http://www.w3.org/RDF for RDF) and dc:subject Resources defined by blog owners. This allows for the dc:subject ontology to be defined by blog owners. Using OWL's owl:sameAs
planetrdf.com (or even the blog owner) could compile mappings between the two ontologies.
Anyway, I hope this generates some discussion or action. It seems like we have the pieces needed to use semweb technologies to make planetrdf.com much smarter. Do others think this is possible?
Also, is it possible to create mappings with OWL based on Literals (for instance, saying that "Blogging" owl:sameAs "blogging")?
My problem with planetrdf.com is that it syndicates all posts from blogs it has chosen, instead of those posts that deal only with RDF and the semweb. For instance, planetrdf.com might show posts on RDF and someone's new car. While I might be interested in that new car (especially if it's a Mini), I would like planetrdf.com to filter out items that do not have anything to do with the semweb or RDF. For a site that is all about RDF, I think we have all the tools necessary.
It should be straightforward for planetrdf.com to say, "This item (blog post) has something to do with one of the categories I syndicate. Let's show it.". Also, it should be just as easy to say "This item is from a category that I do not syndicate, so I'll silently drop it".
In RSS 1.0 (RDF) there should be a triple to indicate an item's category. From the implementations I've seen, the Object in this triple is a Literal (string). For instance, here is a triple you might get from a RSS 1.0 feed:
urn:example:blogitem:xxx dc:subject "Blogging" .
The most simple thing to do here is to create a list of approved Literal values of categories. The planetrdf.com aggregator would simply check the string values of the dc:subject.
Now, of course, we are using Semantic Web technologies, so we can do a bit better. The property dc:subject doesn't specify a rdfs:range. I'm not sure exactly what this means, but I'm hoping it means we could place a Resource as the Object of that triple.
For instance, the above triple could change to:
urn:example:blogitem:xxx dc:subject category:blogging .
Then, using OWL, we can provide mappings between Resources that planetrdf.com is interested in (such as http://www.w3.org/RDF for RDF) and dc:subject Resources defined by blog owners. This allows for the dc:subject ontology to be defined by blog owners. Using OWL's owl:sameAs
planetrdf.com (or even the blog owner) could compile mappings between the two ontologies.
Anyway, I hope this generates some discussion or action. It seems like we have the pieces needed to use semweb technologies to make planetrdf.com much smarter. Do others think this is possible?
Also, is it possible to create mappings with OWL based on Literals (for instance, saying that "Blogging" owl:sameAs "blogging")?