For a position paper, it seems fine to me to put things a little provocative; and, the quote from the TBL SemWeb paper (”without needing artificial intelligence”) is correct, isn’t it?!

They appropriately distinguish between the syntax & tools of the SW, and the next step of semantic interpretation, and just raise a question how much the additional syntax obtained (syntax that hints at semantics) would help data analysis in comparison to not having that syntax. (At least, that is how I interpret it.) I think that is a valid question, and yet to be answered.

Greetings

I think the critique is at least partly valid. Most vocal proponents of semantic web are not doing much to apply really interesting (statistical) methods to it.

And wast majority of (government) research resources go into structured approaches, even though they are hitting into the wall all the time.

People involved in LOD are great guys. But I think LOD has yet to deliver in terms of real utility. Can you name companies doing useful stuff with statistical methods over LOD?

So the challenge is the following: the data is out there (thanks to enormous efforts of semweb proponents). now we need to create really useful services on top of it and there’s not enough of effort applied at the moment.

bye
Andraz Tori, Zemanta

There is another fallacy in the paper, and a well known one, namely equating the Semantic Web with (indeed, possibly expensive) ontologies. If I use Calais or Zemanta that will add URI-s as tags drawn from DBPedia (or similar) I do Semantic Web, and I do not necessarily use complex ontologies. Ontologies are for the Semantic Web (when needed), but the Semantic Web is not for ontologies:-)

I can fully understand that Google wants to exploit the formidable amount of data they harvest, and that therefore statistical approaches play an important role in their research activities. No problem with that. But generalizations and, primarily, unnecessary controversies is not what the Web community at large needs…