Throwing off Ballast (Part I)

As my interests shift I will have to jettison a few things.

My first (innocent) victim is the Semantic Web course I gave the last years at the local Technical University.

The last iteration was definitely smoother than the one before, both in terms of content and organisation. And dealing with students was quite fun. They also always come up with new things and are a good pool to hire from.

But the time to be spent on such course is still exorbitant.

  • Per lecture day I need at least 1 full day preparation, selecting different portions of slides. And we talk about 12 - 13 lectures.
  • Per student assignment (we had 2) I need at least a couple of days to come up with a decent, worked-out example. One which is not completely artificial, but still is complex enough to demonstrate the issues involved.
  • Per assignment it takes 3 days to examine and assess them. And there is also a supplementary day for the sick and late.
  • Per written exam it takes me 3 days to produce a completely new one, less of course if I rehash earlier questions. Marking them takes me another day.
  • And in total I have to provide 4 exams.
  • And you have to count in email support for students and other administrative work.

You do the maths.

Admittedly, you learn most when you teach. Because only then you really make sure that you understand your material when you turn up at a lecture. The weekly exam.

But the Semantic Web does not interest me that much, actually. I also somewhat lost confidence in the usefulness of OWL and SPARQL, or both. Still, I definitely believe that there is room for other semantic technologies.

I wonder whether I will miss standing in class. Pah, probably not for some time.

Posted In

SPARQL

I think SPARQL is kindof cool, have you tried
http://dbpedia.org/snorql/
, lets you do SPARQL queries in DBPedia.

Inge Henriksen (not verified) | Sat, 02/28/2009 - 14:24

Re: SPARQL

I think SPARQL is kindof cool ... DBPedia.

DBPedia is cool.

Ad SPARQL: Ever wondered how you emulate a NOT in the language?

rho | Sat, 02/28/2009 - 14:43

Ad emulating NOT EXISTS in SPARQL

With something like:
...
OPTIONAL { $X :myProperty $Y . }
FILTER( !bound($Y) )
...
:-)

BTW, I think it's really a pity for students who won't be able to attend this course held by you anymore then, also since I found it to be the best one among the ones I did passing winter term (if not even one of the best ones of all I've experienced so far).

Thx & all the best,
Robert

robi42 (not verified) | Sat, 02/28/2009 - 15:14

Re: Ad emulating NOT EXISTS in SPARQL

OPTIONAL { $X :myProperty $Y . } ... FILTER( !bound($Y) )

:-)))) Yes.

... also since I found it to be the best one among the ones I did passing winter term ....

Ok, I would have to lie would I not tremendously appreciate such a compliment. /Me bows slightly.

rho | Sat, 02/28/2009 - 17:39

I just want to confirm

I just want to confirm robi's posting. A pity you don't hold courses anymore. Thanks for the informatory 'semantic web' course and your great sense of humor!
Do you reveal your new research focus to the interested audience?

All the best,
Mat

mat (not verified) | Wed, 03/04/2009 - 15:57

Re: I just want to confirm

I just want to confirm robi's posting. ...

Ok, one more positive feedback, and people will think this is all faked.

Do you reveal your new research focus to the interested audience?

There is nothing secret about it, and it is not a secret that I have been sceptical about the traditional vision of the semantic web. But I certainly think that semantic technologies have a place in the solution stack, albeit not as a stand-alone selling point.

So what I am trying to do (and made some advances already last year) is to conciliate semantic aware applications with the RESTful service style. That may even culminate in a semantic programming language. Which sounds silly, as programming is all about semantics (ok, maybe not in Java). But I think you know where I'm heading. And I'm not alone on that vector.

The other axis I am following is to merge Topic Maps (or actually my extension of it) with text mining techniques (SOMs, SVMs, CSAs, etc.). That together with the above and with some ideas I stole from Robert Cerny might give me a nice appdev environment.

rho | Thu, 03/05/2009 - 14:33