RE: Semantic content negotiation (was Re: expectations of vocabulary)

On Wed, 26 Jul 2006, Miller, Michael D (Rosetta) wrote:

> Hi Xiaoshu,
> I think many excellent points and discussions are being made but I'm
> feeling frustrated because, in the 80/20 paradigm (80% is easy to
> implement the last 20% is much harder), these discussions are in the
> 20%, I might even venture that they are in the top 5%.

Hello Michael, your point is well taken and I share the same concern about 
threads that address issues that are 'somewhat' peripheral (not to 
discourage such conversations, mind you).  For example, 
until recently I was (admitedly) ignorant to the whole LSID / URI 
conversation.  Yet from the points made so far I still have yet to come 
across any drop dead issues that would setup a person that decided to go 
with one naming / resolution convention over another for a later disaster.

Perhaps, there are some other nuances that haven't been mentioned or I 
lack an understanding of those that have been mentioned?

However, I'm not as frustrated with the 80% that aren't addressed as much 
as they should, because I get the impression they are 'on the radar' of 
recently created Semantic Web Deployment (SWD) group, particularly because 
the charter addresses (in my mind) the best way to address this deficiency
80%: well documented best practices.

> The vast majority of the potential consumers (the 80%) of the semantic
> web are just the group I was pointing out, normal researchers who don't
> care how google works, or http, they just use it.  What they would want
> to get from the semantic web is probably out there already and if the
> infrastructure of the semantic web could be set up to reach the already
> existing resources (GO, MO, NCI metathesaurus, etc) in even an
> admittedly limited fashion, adoption and additional resources would
> become available for the semantic web.
> There seems lately in these discussions to be an emphasis on making the
> semantic web useable for the 5% who care about transitive closure and
> for perfect modularity.  These are great things, but they will come
> faster, I believe, if the semantic web becomes available with what we
> have now.
> I've seen relatively little discussion that targets this 80% that is
> available right now, warts and all.

I've believed (for some time) that the main issue is that the literature 
and general area of research upon which Semantic Web Technologies are 
based on (Artificial Intelligence, Knowledge Representation, Expert 
Systems, Logical programming / inference, etc..) are *very* inaccessible 
to the 'everyday' person.

For example, In my experience, in order to find anything on common pitfalls of ontology 
engineering (history, consequences on reasoning performance, 
restrictions on expressiveness, etc.), you have to wade through academic papers with more references 
to logic notation and nomenclature than the everyday Joe Schmoe should 
have to contend with.

I think, well documented best practices that attempt to summarize what is 
already in available literature but very inaccessible is what will address 
this 80%.  As well as best practices for concerns that don't have any 
precendent: Take the document on best practices for deployment of 
vocabularies over HTTP [1] for example.

The existing Working Drafts and Working Group notes [2] is exactly what 
needs to be built on for advocacy.

What I think could be improved upon to address the 80%:

I. Best practices, lessons learned in programming APIs for RDF/OWL

As an implementer of RDF libraries, I've found myself reinventing the 
same wheel that similar implementations in other platforms / languages 
most likely have already faced but for which there are no good references 
for the next victim:

Whats the best way to fashion an API for named graph support, inference, 
query dispatching, etc..

II. Best practices, lessons learned in ontology engineering: Consequences 
of decidability, when to use rules or when to rely soley on DL semantics, 
when is reasoning *really* neccessary, what are some common patterns in 
modelling ontologies

III.  Best practices and common patterns in querying RDF content: common 
kinds of queries, caveats in query processing, how and when to rely on 
querying alone and when to rely on rules, limitations with current query 
languages and their implementations, etc..

IV.  Conventions for consumers of distributed RDF content (i.e., the 
current thread on linking terms and the mechanisms that software 
agents that consume RDF content should follow and assumptions they can make).

Just to name a few.  These kinds of things fall directly under the SWBPD 
charter [3], so I do think the 80% is 'on the radar' but definately not 
adequately addressed, at the moment

My $0.02


Chimezie Ogbuji
Lead Systems Analyst
Thoracic and Cardiovascular Surgery
Cleveland Clinic Foundation
9500 Euclid Avenue/ W26
Cleveland, Ohio 44195
Office: (216)444-8593

> cheers,
> Michael
>> -----Original Message-----
>> From:
>> [] On Behalf Of
>> Xiaoshu Wang
>> Sent: Monday, July 24, 2006 7:10 PM
>> To:; 'Semantic Web'
>> Subject: RE: Semantic content negotiation (was Re:
>> expectations of vocabulary)
>> --Michael,
>>> This group seems to have forgotten that for the semantic web
>>> to be used by more than a handful of hardcore researchers, it
>>> will need have tools that are easy to use for the average joe
>>> researcher.  It feels like there are a lot of levels that
>>> have gotten mixed up in the recent discussions.
>> When I said consumer, I meant it to be those who wrote the
>> software agent.
>> Not the consumer who actually use the agent. Because no
>> matter what, it is
>> the consumer who sends the request info to the provider.  What kind of
>> closure always comes from the consumer.  The issue here is
>> who should be
>> handle it.
>> Xiaoshu

Received on Wednesday, 26 July 2006 17:42:20 UTC