[uf-dev] JSON representation of semantic objects

[Manu Sporny]

Kevin Marks wrote:
> Both those seriliazations look remarkably verbose. Why doesn't:
> 
> <div class="haudio">
>    <span class="fn">Start Wearing Purple</span> by
>    <span class="contributor">Gogol Bordello</span>
>    found on
>    <span class="album">Underdog World Strike</span>
> </div>
> 
> just map to:
> 
> {"fn":"Start Wearing Purple", "contributor":"Gogol Bordello",
> "album":"Underdog World Strike"}

Several reasons:

1. We're going for a format that encapsulates all of the information
that an RDFa, eRDF or Microformats parser can generate. This format is
for developers - not for publishers, designers nor everyday folks.
Developers don't like it when you start chucking away data that should
be accessible to them (such as the encoding language of a data item, or
the type of a data item).

2. RDFa and eRDF can have a variety of vocabularies, of which
Microformats are a semantic subset. We can't boil
"http://purl.org/dc/elements/1.1/title" down to "title" or "fn" for an
RDF semantic object because we lose meaning... which is bad.

3. The serialization approach you give is not scalable - or rather, to
make it scale, we have to come up with another representation that the
eRDF, RDFa and Microformats communities must agree on. It's fairly
straightforward to map Microformats to RDF vocabularies - the opposite
however, is quite difficult.

While the example you give is the simplest representation of a semantic
object, it is not good enough if we want to do an acceptable job of
representing semantic objects generated by the RDFa and eRDF parsers in
Operator (and other applications to follow).

-- manu