telco this Friday, 15:00 (CET)

Dear all,

  first of all, this is a gentle reminder that we will have our regular 
telco this Friday at 15:00 (CET). I am quite happy that this time slot 
seems to suit many people.

There have been some final issues raised that we need to settle. I try 
to summarize them here (apologies for my bad response time in the last 
days, other things have kept me busy!). I think we are at a point were 
we have the issues on the table and simply need to vote on them. I think 
the most important ones are the following three (I list some other minor 
issues below)

1) Strong vs. weak linking

Guido has recommended that we do not create formal links to other 
models, e.g. by equivalentClass or subClassOf axioms, but we use "weak" 
and "informal" links through rdfs:comment and the like.
 From a SW perspective, this approach is non-orthodox I think. The whole 
SW is about formalization. Informal links can not be used by machines to 
align datasets, do distributed querying etc. Informal links in this 
sense are only documentation and for human inspection and reasoning. 
This is exactly what the SW wants to avoid, i.e. that we always have to 
appeal to intuition of people ;-)
I do not have a very strong position on this issue, but if some of us 
would still give a few cents for the SW idea, then we should go for a 
more formal and thus stronger linking. Just my two cents.

2) Lexical Entry rdf:subClassOf semio:Expression?

We all agree that Lexical Entries are lemmas and thus essentialy types 
that can be instantiated in a text, thus becoming a token informally 
speaking.
I have no strong position here as well. If Also says that types in this 
sense can also be semio:Expressions then I am fine.

3) Formalization of range of "reference"

We all agree that reference can have owl:Class, owl:Individual or 
owl:Property (object and datatype as range).
Aldo/Armando have brought in the possiblity of also owl:Ontology to be 
in the range. I do not see the need for the latter honestly. Of course, 
there can be ontologies that axiomatize single sentences, but then the 
lexical entry does not "denote" or "reference" the whole ontology but 
one class or property in that ontology which is axiomatized using a 
number of helper classes and axioms.

We can formalize the range, fine. But let's assume that people mis-use 
our model, they will for sure! They might use all sort of things as 
"references". We would then infer that these things are owl:Class OR 
owl:Individual OR owl:Propery OR owl:Ontology.
First, this inference is quite useless in my view as we are still left 
with uncertainty about the specific status of the entity in question. 
Second, this might have unwanted implications, possibly leading to 
inconsistencies.
I do think that we should follow here the principle of "graceful 
degradation" in the sense that the system wil not break if you mis-use 
it a bit. Actually, to close the cycle, I just that this same argument 
could be used to argue against more formal links in general (see my 
point 1). I apologie for this inconsistency in my thinking. I wish I 
could indeed always consistent with myself ;-)

On the telco on Friday, I will call for votes on these three isuess, 
allowing anyone to summarize their position at the beginning of the 
discussion of each of these points.

An email summarizing other minor issues will follow:

Best regards,

Philipp.


-- 
Prof. Dr. Philipp Cimiano
Semantic Computing Group
Excellence Cluster - Cognitive Interaction Technology (CITEC)
University of Bielefeld

Phone: +49 521 106 12249
Fax: +49 521 106 12412
Mail: cimiano@cit-ec.uni-bielefeld.de

Room H-127
Morgenbreede 39
33615 Bielefeld

Received on Thursday, 18 July 2013 06:50:47 UTC