On Wed, October 24, 2012 06:36, Aldo Gangemi wrote:
> ... we have realized that in some cases systematic
> polysemy (as James Pustejovsky baptised it, cf. [6]) of e.g. Stadium as a
> facility and a place, or some of the examples in this thread, is somehow
> "ineliminable", because the intended conceptualization cannot be reduced
> to a specific concept, but comes packed with a frame (or knowledge pattern
> as I have discussed in [4]), in which the same term is used in multiple,
> metonymical ways. (01)
Exactly. 15 years ago, Cyc was dealing with the organization/place
dichotomy, having the general terms #$GeographicalAgent and
#$GeographicalRegion. What Cycorp did was to create a context
in which #$GeographicalAgent was a subclass of #$GeographicalRegion
and another context in which #$Agents were disjoint from #$Places. (02)
If one needed to reason with polysemy, the first context (called a "dualist"
context) was used, if one needed to keep the distinction, the second
context was used, otherwise, the issue was not forced. A function
(#$TerritoryFn) was defined to refer to the region given the agent.
An excerpt from the comment:
"The function #$TerritoryFn, applied to some instance
GEO-AGENT of #$GeographicalAgent, denotes the spatial region
of the earth occupied by GEO-AGENT, usually demarcated by
natural or artificial boundaries. Sometimes these boundaries are
determined by a political process (as in the case that GEO-AGENT
is a #$GeopoliticalEntity). Note that this geographical region can
be fluid over time, expanding and contacting as the geographical
agent's borders change. Note also that the temporal extent of
(#$TerritoryFn GEO-AGENT) is equal to the temporal extent of
GEO-AGENT."
[FWIW, there's a slight error in this comment, in that in a context in
which GEO-AGENT would have territory on some extra-terrestrial body,
#$TerritoryFn would not denote "a spatial region of the earth".] (03)
> This bare fact of human cognition and life appears also in more
> regimented domains, for example in medicine I noticed several
> times (e.g. [5]) the key role played by ambiguous notions that
> activate a complex frame, and cannot be broken into pieces for
> any plausible interpretation, e.g. "inflammation" or "diagnosis". (04)
Inflammation is a good example of another standard dichotomy --
that of an event and the "doer" of the event. A more standard
example often given is StormAsEvent vs. StormAsObject. For
some cases, it might be considered useful to tease such distinctions
apart, but for others, maintaining the ambiguity can be better.
If one is keeping the ambiguity, one might need to create new
varieties of some general properties to cover the ambiguous case. (05)
> In general, I'd say that ontology engineering should shift towards an
> empirical approach that validates and motivates the application of formal
> theories, logic, and philosophy in practical applications. (06)
Agreed. That does not mean that appropriate distinctions should not
be made in the correct contexts or that rules for both validation and
for deriving additional information might not be useful. (07)
-- doug f (08)
> Ciao
> Aldo
>
> [1] http://www.sciencedirect.com/science/article/pii/S0010027711002496
> [2] http://wit.istc.cnr.it/stlab-tools/tipalo
> [3] http://iswc2012.semanticweb.org/sites/default/files/76490065.pdf
> [4]
>
>http://iospress.metapress.com/content/6038458357588v43/?p=06bb61df7d6a43728a171d8d5510284a&pi=9
> [5] Gangemi A, Pisanelli DM, Steve G, Understanding Systematic
> Structures in Polysemous Medical Terms, in: J Marc Overhage (ed.),
> Conceptual Converging Information, Technology, and Health Care,
> Proceedings of the 2000 AMIA Annual Symposium.
> [6] http://www.aclweb.org/anthology/J91-4003
>
>
> On Oct 24, 2012, at 11:24 AM, Martin Hepp
> <martin.hepp@xxxxxxxxxxxxxxxxxxx> wrote:
>
>> Dear all:
>> Since the headline of this discussion is still my EKAW 2012 talk ;-)
>>
>> One of my main point in the talk was that Web ontologies sit somewhere
>> between (many) people and machines, so their type system must try to
>> balance out conflicting requirements, for instance what I called the
>> "degree of disambiguation / discriminatory value of a type" vs. the
>> reliability of type membership.
>>
>> In general, types that provide fine-grained distinctions are more
>> valuable for automated processing, like
>>
>> book - > a) book copy, b) book title
>>
>> restaurant -> a) legal entity b) location
>>
>> etc.
>>
>> However, the quality of data (to be more precise: the reliability of
>> type membership information) can suffer from more subtle distinctions in
>> the type system, because human users who are publishing data, either by
>> explicit annotation or by writing mappings to legacy data structures,
>> may not be able to understand and/or apply the conceptual distinctions.
>>
>> So we cannot take it for granted that an ontologically better type
>> system with cleaner distinctions enhances interoperability and reuse of
>> data.
>>
>>> From the adoption of GoodRelations, we see that some of the more
>>> valuable distinctions, in particular product vs. product model and
>>> business entity vs. location, are often applied incorrectly by the
>>> owners of data.
>>
>> So when we look for types for Web ontologies, we must optimize them so
>> that we find the most effective ratio between
>>
>> a) the reliability of the type membership at scale and
>> b) the discriminatory value of the types.
>>
>> That is an open research question.
>>
>> By the way, I am not saying that this holds for all ontologies. But I am
>> convinced it is one of the key challenges of building **Web**
>> ontologies, and since so many people take the Web as a motivation for
>> ontology research, I think this is an important point to make.
>>
>> Best wishes
>>
>> Martin Hepp
>>
>>
>> On Oct 24, 2012, at 5:46 AM, John F Sowa wrote:
>>
>>> On 10/23/2012 6:59 PM, William Frank wrote:
>>>> I find it troubling that a /place/ is a kind of thing, yet
>>>> a/date/time/
>>>> is a "data type," and not a kind of thing.
>>>
>>> Space and time are the containers that hold everything in the universe.
>>> A region in space or an interval of time could be considered physical.
>>> In fact, quantum mechanics implies that even a region of outer space
>>> that is a perfect vacuum would still have a "zero-point energy".
>>>
>>> But the coordinates of a location in space or time would be data.
>>> I would classify dates and times in the same way as geographical
>>> coordinates.
>>>
>>> In any case, I prefer the term 'entity' for the most general category
>>> at the top of the hierarchy. Literally, an entity is anything that
>>> exists or might exist. In CLIF, you could say
>>>
>>> (forall (x) (entity x))
>>>
>>> In my preferred ontology, I would say that events exist, words exist,
>>> relations exist, patches of colors exist, and information exists.
>>> The term 'thing' sounds too concrete for them. So I prefer to
>>> call them all entities.
>>>
>>> John
>>>
>>> _________________________________________________________________
>>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>>> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
>>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>>> Shared Files: http://ontolog.cim3.net/file/
>>> Community Wiki: http://ontolog.cim3.net/wiki/
>>> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>>>
>>
>> --------------------------------------------------------
>> martin hepp
>> e-business & web science research group
>> universitaet der bundeswehr muenchen
>>
>> e-mail: hepp@xxxxxxxxxxxxxxxxxxx
>> phone: +49-(0)89-6004-4217
>> fax: +49-(0)89-6004-4620
>> www: http://www.unibw.de/ebusiness/ (group)
>> http://www.heppnetz.de/ (personal)
>> skype: mfhepp
>> twitter: mfhepp
>>
>> Check out GoodRelations for E-Commerce on the Web of Linked Data!
>> =================================================================
>> * Project Main Page: http://purl.org/goodrelations/
>>
>>
>>
>>
>> _________________________________________________________________
>> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
>> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
>> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
>> Shared Files: http://ontolog.cim3.net/file/
>> Community Wiki: http://ontolog.cim3.net/wiki/
>> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>>
>
>
> _____________________________________
>
> Aldo Gangemi
>
> Full professor
> Department of Informatics (LIPN)
> Paris Nord University - Sorbonne Cité - CNRS
> 99, av. J-B. Clement
> 93430 Villetaneuse France
> aldo.gangemi@xxxxxxxxxxxxxxxxxxxx
>
> Associate Researcher
> Semantic Technology Lab (STLab)
> Institute for Cognitive Science and Technology,
> National Research Council (ISTC-CNR)
> Via Nomentana 56, 00161, Roma, Italy
> Tel: +390644161535
> Fax: +390644161513
> aldo.gangemi@xxxxxx
> http://www.stlab.istc.cnr.it
> http://www.istc.cnr.it/people/aldo-gangemi
> skype aldogangemi
> okkam ID: http://www.okkam.org/entity/ok200707031186131660596
>
> _________________________________________________________________
> Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
> Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
> Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
> Shared Files: http://ontolog.cim3.net/file/
> Community Wiki: http://ontolog.cim3.net/wiki/
> To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
>
> (09)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J (010)
|