Re: [hcls] The future of Semantic Web / Linked Data R&D; translating research into practical applications (was: Updated wiki page for HCLS Knowledge Base)

Hi Matthias!

I surrender to all your arguments below, and to those of Michel also :-)   
I think we are simply talking from different perspectives and  
life/career-situations that have different immediate needs.

I don't do "biological research" per se, so for me a warehouse  
(semantically-enabled or not) is not a useful end-point; however, I can  
see how it might be useful, and might be the FODDER for research,  
especially for those interested in answering biological questions.  My  
problem is that I have to justify - to funding bodies, to my department  
head, to my tenure-review committee, and to my external reviewers - how  
this semantic web "stuff" is in any way research!  This becomes a very  
difficult case to make when the most obvious outcomes from the community  
to date are (a) ontologies used for annotation, and (b) warehouses.  It's  
hard to argue against a reviewer who comments that, as far as they can  
tell, the purpose of the semantic web is to come up with a fancy and  
expensive way to construct a database!

Anyway, I don't see that I have anything more useful to say on the  
subject.  The counter-arguments are valid, and I believe that my  
perspective is also valid.  The two "sides" (though I think we're all on  
the same "side" :-) ) simply have different end-requirements for engaging  
in the activity in the first place.  I'm just being grumpy about the issue  
due to personal challenges related to the topic we are discussing :-)

Cheers all!

Mark




On Wed, 14 Oct 2009 06:55:43 -0700, Matthias Samwald <samwald@gmx.at>  
wrote:

> Hi Mark,
>
>> Moreover, warehousing in and of itself isn't research, nor is it pushing
>> "the state of the art",
>
> I have become a bit weary of this interpretation of 'pushing the state of
> the art' in this context. Looking at the set of standards, datasets,  
> tools,
> practices around RDF/OWL that we now have available, I think that the
> Semantic Web community has made astonishing accomplishments in the last
> decade. These things open up great possibilities that most people  
> outside of
> the community have not even become aware of.
> I would be glad to see more researchers in the community directing their
> research  towards APPLYING this existing set of excellent technologies to
> problems in the real world, rather than solely 'pushing the state of the
> art' regarding the underlying technology. This does not mean that there  
> is
> no place for basic research anymore -- there are many important and
> intriguing research questions around usability, the development of
> domain-specific vocabularies / ontologies, the specific needs and
> requirements of certain domains, et cetera.
>
> I also wish that funding agencies would do more towards encouraging the
> translation of results from basic Semantic Web / Linked Data research  
> into
> practical applications. There would be many exciting questions for basic
> research hidden in that research program, and it might help to increase  
> the
> positive impact of our work so far.
>
>> We don't need RDF to make warehouses!
>
> RDF enables us to do many things. Among these things is the possibility  
> to
> aggregate different datasets into a single triplestore, where the  
> datasets
> are integrated ad-hoc by virtue of shared identifiers, taxonomies and
> ontologies. If needed, parts of the triplestore can be exported as RDF,
> integrated with other data, re-purposed et cetera with great ease. This  
> is
> very different from a classical "data warehouse" (as I understand it),  
> where
> a lot of code has to be written and manual integration work has to be  
> done,
> and where it is very difficult to re-purpose, 'mix and mash' data once  
> it is
> inside the warehouse.
>
>> This is why I  am "irked" when I see our own group admitting that
>> RDF-Web-crawling is  "nice", but it's just too much fuss, so we'll
>> capitulate and build a  warehouse anyway.
>
> Again, I think you are introducing a false dichotomy. There is no
> 'capitulation' here, nor is it making a choice between two mutually
> exclusive options. We can make both things part of our toolset, and  
> select
> the right tool for each task that we are facing. Furthermore, we are not
> building a 'warehouse', as described above.
>
> Have a nice time in Banff,
> Matthias
>

Received on Wednesday, 14 October 2009 14:42:23 UTC