Talk:Semantic Sidebar

From GetSemantic

[edit] Why use Semantic Markup

I don't completely agree with the approach of this page. I think we ought to encourage semantic markup for a purpose, not simply for its own sake.

Because there are quite a few different ways in which markup can be made more semantic, it is important to know what someone wants to achieve with their markup before recommending methods.

[edit] When to use what technique

I disagree with the blanket approach of using microformats wherever you can, and eRDF/RDFa for the rest.

I think the correct approach depends on what the author/developer wants their markup to do:

If you want to be able to perform a simple transformation from the html information into another format (eg: vcf, ical, Atom, DocBook, etc etc), then look for a microformat (because that will probably come with a stylesheet already written), and if one doesn't exist, make up your own convention based on the format and write a stylesheet or script to do the transformation.

If you want to target specific tools and aggregators (eg: Operator, Technorati), then obviously use the convention that they recognise (eg: rel-tag, rel-nofollow).

If you want to describe your information so that it can be transformed into RDF and smooshed up, mashed up, merged and migrated, then choose between eRDF, RDFa and GRDDL. With eRDF and RDFa, it comes down largely to what tools you want to use, and which syntax fits best with your personal preferences and authoring requirements. With GRDDL, syntax isn't an issue, and it can be a solution if you need a non-RDF based brand of semantic html for the functional reasons mentioned above (or have legacy markup amenable to transformation), but also want to make your data available as RDF.

That's my take on it anyway - and for me, asking "what do you want to do with it?" really helps clarify the choice of technique, and hopefully sidesteps most of the flamewar territory

[edit] In defence of Reckless Ontology Creation

I don't think it's a good idea to make RDF seem harder or more formal than it has to be.

Again, it comes down to purpose. If interoperability is important, then reusing existing ontologies - the ones used by the other tools and datasets you want to use your data with - is the right course, and will pay dividends.

On the other hand, if all you really want/need is to get your data into triples, then it doesn't really matter.

We need to promote more of a "(my) machine first, (other) people second" approach to RDF if we want people to publish more of it.

--keith