Re: Where are the Linked Data Driven Smart Agents (Bots) ?

On Thu, Jul 7, 2016 at 12:51 PM, Ruben Verborgh <ruben.verborgh@ugent.be>
wrote:

> Hi Juan,
>
> Seems like we mostly agree—short remarks below.
>
> > One thing is science. Another is engineering.
>
> Perhaps we need Semantic Web Engineering conferences then as well!
>

That's why you have developer conferences, etc. ISWC this year had a
Applications track. And it doesn't have to be semantic web specific.


>
> > If we don't know the right evaluation metrics (I agree with you that we
> don't), then that is the current challenge we, as a semantic web scientific
> community, have to tackle.
>
> Indeed, but I've found the scientific community to be not so open to new
> evaluation metrics either. There is insufficient agreement on (and too
> limited knowledge of) the right scientific methodology to tackle such novel
> problems.
>

Well, I don't know what you have proposed. But it seems that you haven't
made a convincing argument :P (if you want to discuss offline what you are
doing, send me an email)


>
> > It shouldn't discourage you... on the contrary, it should encourage you
> to identify novel ways to evaluate what you are doing and convince the
> community why it is important.
>
> The trouble is you don't have to convince the entire community (with whom
> you can have an open dialog), but a tiny set of anonymous reviewers (for
> whom the known paths are often easier to judge).
>

Paper reviewing... that is different topic I don't want to get into here.


> My remark was precisely that convincing is hard once you move away from
> the known paths.
>
> So the scientific community, which is a large part of the total Semantic
> Web community, might in that sense be hampering real novelty—from science
> and engineering alike, whichever might be the difference.
>

There is a clear difference between Science and Engineering. That is my
point.  Science is about understanding what is unknown. In this case, it is
not well known how to evaluate new types of systems. That is what needs to
be studied. We need to figure out how to evaluate and evaluation. It is a
bit meta. Jim Hendler stated this 7 years ago and to the best of my
knowledge, this is an area that hasn't been tackled (PhD thesis anybody?)

If your goal is to get something out there and for it to be used, then why
do you bother spending time publishing papers. Look at all the open source
projects changing the world, with very little to no scientific
publications.

Just do what makes you happy and be the best at it. Strive for excellence!
I know you are :)


> Best,
>
> Ruben