Two comments to the FAQ

All,

After intense enquiries, I had a look in the FAQ. I think it is mostly 
fine, good enough to publish, FAQs shouldn't be too big from the 
outset, it isn't QWWYA (Questions We Wished You'd Ask), it should be 
responses to actual things that people are often curious about.

I have two comments, one is about the relationship to microformats, 
http://www.w3.org/2001/sw/SW-FAQ#Folskonomi [sic!]

where it is stated that ontologies are usually larger and more rigorous. 
I think that is not necessarily true. To the contrary, it looks 
like "if it's not on microformats.org it's not a microformat", whereas 
anyone who has a URI space can create an ontology. Like my first 
gallery ontology, http://my.opera.com/community/xmlns/2005/gallery.rdf 
was created without asking anyone, whereas version 2 of it 
http://my.opera.com/community/xmlns/2006/gallery.rdf
was created after two hours on IRC. I think we're more decentralised and 
flexible than most, and that we should emphasize this.

Now, I haven't been participating too much in the microformats 
community, I can't really say what distinguishes us, but let me just 
throw this out: With microformats you need programmer interference for 
everything, even writing the (simple) parser. You can't connect the 
data trivially, and you have a namespacing problem. With RDF you have a 
data model, you have a query language. I think we need more 
clarification on this topic. 

My other comment is regarding tools:

http://www.w3.org/2001/sw/SW-FAQ#tools

I think it is hard to say that our tools are of the same quality that 
XML tools are. For example, Robin Berjon, one of the great Joost 
hackers, posted a message to the new Perl-RDF list 
http://lists.perlrdf.org/pipermail/dev/2007-February/000004.html
saying 
"I've long been interested in RDF, but have long been frustrated  
with the quality of the tools (not just in Perl, the Java ones are  
far from ideal either) and as a consequence I'm still dabbling with  
using XML + XQuery instead of RDF + SPARQL for the areas that  
interest me personally"

I've personally spoken with a lot of people who see the performance of 
tools that are available as really bad, including some really good 
hackers. This, IIRC, was also said in response to our enterprise 
questionnaire. 

There is a real risk over overstating our case here, and make early 
adopters turn around.

OTOH, we see that some have 1.5 billion triples, we see successful 
large-scale enterprise deployments, and think it would be very 
interesting to learn about what they've done right, and if they have 
bumped into the performance issues others have had and how they 
overcame them.

Cheers,

Kjetil
-- 
Kjetil Kjernsmo
Semantic Web Specialist
Opera Software ASA

Received on Wednesday, 28 February 2007 17:02:51 UTC