[Top] [All Lists]

[ontolog-forum] Re: questions to Ontolog Invited Speakers - Mr. Jack Pa

To: jakubkotowski@xxxxxxx
Cc: jack.park@xxxxxxx
Cc: ontolog-forum@xxxxxxxxxxxxxxxx
From: Patrick Durusau <patrick@xxxxxxxxxxx>
Date: Thu, 27 Apr 2006 18:39:46 -0400
Message-id: <44514832.8060205@xxxxxxxxxxx>
Jakub,    (01)

Excellent questions!    (02)

One quick point: The focus of the presentation was on federation of 
ontologies and quite naturally we focused on merging subjects (concepts) 
that you would find in an ontology.    (03)

HOWEVER, the same principles apply for example to subjects that you wish 
to say are an instanceOf a concept in an ontology (standing alone or 
merged with another).    (04)

For example, assume we have the subject (concept) automobile in an 
ontology. There are subjects, possibly on a car dealer's lot, that we 
wish to say are instancesOf that subject. We could simply list all the 
information that is on the "sticker" of each for each subject, but not 
all of that information is relevant to saying any one of them is an 
instanceOf automobile. From a subject map perspective, we would identify 
the properties of each that we think qualify it as an instanceOf that 
subject.    (05)

Jakub Kotowski wrote:    (06)

>I have several questions to the speakers if I may:
>1. On the last slide you mentioned that subject maps enable ontological 
>reasoning even in the absence of data being formally “ontologized.” Could 
>please explain more what you mean by that? I am particurarly interested how 
>the proxies are semantically interpreted - in terms of description logic 
>perhaps. Do they even have such semantic when they don't have a syntax and 
>structure? If not, how is the reasoning possible then?
In other words, properties are what identify the subjects we wish to 
reason about. That is to say that having formulated an ontology, no mean 
task, we want to use it, but that requires us to identify subjects that 
we think match the subjects (concepts) in our ontology.    (07)

Since we have to define what those properties are, so that we know a 
subject (for merging, reasoning, etc.), if we encounter those properties 
in a data set, we have found that particular subject, whether an 
ontology has been formally applied to it or not.    (08)

It is making the properties explicit and declaring what properties 
identify which subjects, that enable us to then reason about a subject 
(concept), whether it has been the subject of formal ontological 
analysis or not. Having identified it, what more need I do to reason 
about it?    (09)

>2. What would happen if one of the merged ontologies would be contradictory? 
>Would that contradiction automatically transfer to the other ontologies 
>during reasoning? I mean really a contradiction in one ontology not 
>contradicting terms/concepts between two ontologies.
Hmmm, no. Let's assume that we have fully merged two ontologies and one 
has a contradiction and the other does not. If we use the ontology 
without the contradition, then its reasoning, since it only uses "its" 
ontology, would produce no contradiction. If we use the "bad" ontology, 
the one with a contradiction, then yes, we would run into a contradiction.    (010)

We are no worse off than we were with the separate ontologies with one 
important difference. If we compare the reasoning results of both 
ontologies, which operate independently of each other, then we have a 
basis on which to notice the contradiction. That is to say that we get 
different results depending upon the ontology we have chosen to use.    (011)

So to that extent, I think we are better off since there is some basis 
on which to see the contradiction.    (012)

>3. How do the subject maps relate to so-called Context OWL (C-OWL) (ISWC 2003, 
>LCNS 2870, pp. 164-179, 2003).
I was only quickly able to find an extended abstract for that article 
which says:    (013)

> C-OWL defines the following kinds of bridge rules stating that a 
> concept from
> an ontology Oi is more general, more specific, equivalent, disjoint or 
> overlapping with a
> concept from another ontology Oj :    (014)

Subject maps are more general as there is no restriction on the 
properties you can use to identify a subject nor are there any 
restrictions on the rules for merging (or not merging) two subject proxies.    (015)

Note that as we pointed out in the presentation, sameAs isn't terribly 
helpful as we don't know on what basis something has been seen as 
sameAs. Does it mean to match all the properties that have been 
accumulated? More or less than half of all the properties?    (016)

The problem is one of opacity, that is to say that the system has no 
capacity to represent the information relied upon by the ontologist or 
user in saying that two subjects (concepts) are the same.    (017)

Thanks for the questions!    (018)

Hope you are having a great day!    (019)

Patrick    (020)

>Thank you,
>Jakub Kotowski
>    (021)

Patrick Durusau
Chair, V1 - Text Processing: Office and Publishing Systems Interface
Co-Editor, ISO 13250, Topic Maps -- Reference Model
Member, Text Encoding Initiative Board of Directors, 2003-2005    (022)

Topic Maps: Human, not artificial, intelligence at work!     (023)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (024)

<Prev in Thread] Current Thread [Next in Thread>