W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > June 2007

Re: Evidence

From: CNR-ISTC <aldo.gangemi@istc.cnr.it>
Date: Fri, 22 Jun 2007 00:24:25 +0200
Message-Id: <429EC599-D2A2-4A90-A31F-E2F833EE7198@istc.cnr.it>
Cc: CNR-ISTC <aldo.gangemi@istc.cnr.it>, public-semweb-lifesci@w3.org, alanruttenberg@gmail.com, samwald@gmx.at, Dan Brickley <danbri@danbri.org>
To: Pat Hayes <phayes@ihmc.us>
Thanks for the last word, Pat, but I think we basically agree,  
although there are minor issues, see comments inside

Il giorno 21/giu/07, alle ore 23:41, Pat Hayes ha scritto:

>> Hi all, Dan (Brickley) pointed me at these discussions on DOLCE.
>
> Well, actually I only mentioned BFO. But you are correct in  
> surmising that I had DOLCE partly in mind.
>
>> I  tend to avoid too many emails and forums on foundational  
>> ontologies, in favor of constructive work, but this mailing list  
>> seems oriented at doing real things, then I am eager to come back  
>> to my old home in medical ontologies :)
>>
>> Some comments and recap on the current status of DOLCE and its  
>> utility for medical ontologies.
>>
>> (1) Continuant vs. Occurrent
>>
>> Pat, we already had this discussion at least two times; the last  
>> one was very detailed and we (apparently) agreed on the fact that  
>> we might be free to adopt the distinction or not, even in DOLCE:  
>> if you want to stay neutral, just use Entity or SpatioTemporalEntity.
>
> Yes, I know you can be 'neutral' in DOLCE. But this isn't using  
> DOLCE correctly, as it was meant to be used. And I was intending to  
> make a stronger point: not just that one can be continuant/ 
> occurrent neutral, but that having the distinction in a high-level  
> ontology at all is actively harmful. The rest of your message makes  
> my main point for me.
>

Maybe, but it would a bad service not to provide the possibility of  
making that distinction at all. In the pattern approach, you can buy  
it or not, and if not, it simply disappears from the ontology, unless  
you want to recover the original context where the pattern has been  
extracted from

>> The distinction can be useful in some domains that share a common  
>> sense, linguistic (Western-variety) intuition, but I will not  
>> fight for it, as for any other.
>
> If only everyone else had your laissez-faire attitude.
>
>
>> The point of DOLCE and related ontologies is having explicit  
>> *rationales* to justify modelling choices, not to dictate how  
>> people should think or model the world. As a matter of fact, in  
>> the context of the NeOn project (<http://www.neon- 
>> project.org>http://www.neon-project.org), we are moving to a  
>> "design pattern" approach to ontology reuse, which will probably  
>> change the way foundational or reference ontologies should be used  
>> or thought about.
>
> That sounds interesting, indeed. Can you point to more on this topic?
>

The original intuition (besides the suggestions from Alexander and  
Gamma, which are not specific to ontology engineering) is from Peter  
Clark ("Knowledge Patterns", KR2000). Then see my paper at ISWC2005,  
and following works. An extended paper of mine will appear on the  
next Handbook on Ontologies edited by Staab and Studer.
Implementation of the repository with an initial set of patterns will  
be published next autumn on the NeOn site, together with a full- 
fledged deliverable.

>> (2) Roles and occurrents
>>
>> Roles, as they are modelled in DOLCE-Lite-Plus (<http://www.loa- 
>> cnr.it/ontologies/DLP_397.owl>http://www.loa-cnr.it/ontologies/ 
>> DLP_397.owl) and in DOLCE-Ultralite (<http://www.loa-cnr.it/ 
>> ontologies/DUL.owl>http://www.loa-cnr.it/ontologies/DUL.owl), are  
>> applied to continuants, but this is just a terminological choice  
>> that adheres to the usual intuition of roles played by agents,  
>> substances, etc.
>
> But this is disingenuous. Maybe it was just a terminological choice  
> when you made it, but now its in the foundational ontology it has  
> the force of case law. It is virtually impossible for a user to  
> deny it: if he or she tries to, all kinds of tools will give error  
> messages, etc.. It really isn't possible to ignore the distinction  
> under these circumstances. One has to be VERY careful when writing  
> things that will get set into stone in a 'standard'.

This is not sustainable, I'm afraid: language and lexicon change,  
ontologists can have headache ... I think we can always provide new  
skins to ontologies, through lighter versions, new versions,  
labelling, extracted patterns.
What we cannot do is establishing "right" names once and forever.

>
>> However, a similar intuition is provided for occurrents by the  
>> classes Course and Task in DOLCE-Lite-Plus and by the classes  
>> EventStructure and Task in DOLCE-Ultralite.
>
> A perfect illustration of my point. Why should anyone have to  
> understand and remember all this, all these complicated  
> distinctions about what goes with what but not with what other?  
> What USE is any of this stuff? What would be lost if it were all  
> simply swept away?
>

The immediate use is giving names to possible distinctions. If names  
are difficult or create confusion, they will not survive, and will be  
changed, or shadowed, or overruled in some way.
BTW, distinguishing between object roles and event roles can be very  
useful in some use cases, for example when you want to express that  
different tasks have been assigned to different roles, independently  
from who is doing what.

> There are entities that occupy space and last for a time. One can  
> use a single, uniform, set of concepts to refer to the spatial and  
> temporal relations between such things, and their mereology and if  
> necessary their geometry and topology, simply *because* they occupy  
> space and time. There is no need to pay attention to all this  
> clutter of terminological distinctions.
>
> Now, of course, there are distinctions to be made: not all  
> spatiotemporal things are in the same category. But they are all  
> *contained in space and time* in the same way, which is why (you  
> wanted a rationale, right?) any of their properties or relations  
> *which arise from or refer to the nature of this containment* can  
> be described in one way, and used for all of them. Which simplifies  
> things enormously and means that busy, practical biologists don't  
> have to keep wondering whether the Krebs cycle or a computer  
> program is a continuant or an occurrent. And then, once that  
> distinction goes away, all these other distinctions go away as well.

If the distinction is useless, go up one level (or use a pattern  
without it), and ignore it

>
>> In general, the class Concept is used to talk of notions that are  
>> used to classify any entity at some time for some reason.
>
> Thats a nice definition. I presume it is OK to parse it as  
> "...classify (any entity at some time) for some reason."

Correct :)

>
>> If you want to use Evidence as a role for (the result, execution  
>> of) an experiment, then you can use EventStructure or directly  
>> Concept. The distinction, where needed, makes sense: roles of  
>> occurrents have usually a richer structure than roles of  
>> continuants, because they are used to suggest how events and their  
>> temporal structure should be interpreted in some context
>
> How do you get that from the usual conditions on continuants? How  
> do contexts come into the picture? Surely events have contexts too,  
> no?

Contexts can be represented as "descriptions" or "situations" in  
DOLCE-Ultralite (coming from D&S ontology), depending on if you want  
to talk about interpretation context or observational context.
An observational context ("situation") can include either objects or  
events, as well as values, qualities, etc. If you do not need object/ 
event distinction, a situation will include generic entities

>
>> ; in order to be an evidence, the result of an experiment should  
>> be obtained in a certain way, e.g. with explicit methods and  
>> control conditions.
>> Pat, notice that in DOLCE-Ultralite you can introduce experiment  
>> results as entities, classified by an "EvidenceConcept"
>
> I would suggest that anything that you can refer to using a name is  
> an "entity" of some kind.

Exactly, and situations can be used as the contexts that "unify"  
those entities when they are classified by that concept (or others).

Ciao
Aldo

>
>> : who cares about Brentano's theology (poor guy, however, did he  
>> make anything bad to your ancestors? :)).
>
> No :-) I got fed up having Barry cite him at me every time we  
> argued, so I went and checked out where his ideas really came from.  
> In fact Brentano is one of the main sources for the 'continuant'  
> notion: both Barry and Peter Simons explicitly acknowledge him. It  
> does not come from robust Western linguistic common sense, as you  
> imply above and many people assume. (It is in fact a *very*  
> peculiar notion.) And Brentano liked the idea because he was  
> absolutely sure that he had a soul, and that souls had to be  
> continuants for basically theological reasons. Makes you just love  
> philosophy, finding out things like this.
>
>
> OK, enough from me on this on this topic on this email list. I  
> promised not to debate this stuff here. So Aldo, you get the last  
> word, if you want it.
>
> Pat
>
>>
>> Something else on another thread about DOLCE and solutions for  
>> readability of horrible foundational terms.
>> Cheers
>> Aldo
>>
>>>  >On Jun 12, 2007, at 3:53 PM, <mailto:samwald@gmx.at?Subject=Re% 
>>> 3A%20Evidence&In-Reply-To=%253Cp06230907c29c6cae1e1a%40% 
>>> 5B10.100.0.39%5D%253E&References=%253Cp06230907c29c6cae1e1a%40% 
>>> 5B10.100.0.39%5D%253E>samwald@gmx.at wrote:
>>>  >
>>>  >>
>>>  >>Hi Waclaw,
>>>  >>
>>>  >>>Matthias, if you look carefully at BFO, you'll see that roles  
>>> are
>>>  >>>entities.  This means that evidences, as roles, are entities.
>>>  >>
>>>  >>Of course. I just wanted to differentiate that an experiment  
>>> is not
>>>  >>an instance of any class called 'evidence' (in other words, an
>>>  >>experiment 'is not' evidence). Instead, it should be associated
>>>  >>with an 'evidence-role'.
>>>  >
>>>  >The only problem with this is that roles inhere in continuants
>>>  >rather than in occurrents. One way around this is not to say that
>>>  >evidence is an experiment, but rather the results of an  
>>> experiment.
>>>
>>> If I may interject, the fact that you need to find a way 'around'
>>> this illustrates what I have long found to be the case, that the
>>> continuant/occurrent distinction, and the resulting artificial
>>> restrictions that it places upon what one is allowed to say, is more
>>> harm than it is worth. One can take any ontology (such as BFO) that
>>> is based up on it and simply erase the distinction (and all its
>>> consequent distinctions) and nothing is thereby lost, only a
>>> simplification achieved and the need for artificial work-arounds
>>> diminished. It is in any case based on very debatable (and indeed
>>> debated) philosophical assumptions, arising chiefly from
>>> ordinary-language philosophy (and Brentano's theology) than from
>>> anything scientific. It carves nature at language's joints rather
>>> than nature's joints.
>>>
>>> Pat Hayes
>>>
>>>
>>> -- 
>>> -------------------------------------------------------------------- 
>>> -
>>> IHMC		(850)434 8903 or (650)494 3973   home
>>> 40 South Alcaniz St.	(850)202 4416   office
>>> Pensacola			(850)202 4440   fax
>>> FL 32502			(850)291 0667    cell
>>> phayesAT-SIGNihmc.us       <http://www.ihmc.us/users/ 
>>> phayes>http://www.ihmc.us/users/phayes
>>>
>>
>> _____________________________________
>>
>> Aldo Gangemi
>>
>> Senior Researcher
>> Laboratory for Applied Ontology
>> Institute for Cognitive Sciences and Technology
>> National Research Council (ISTC-CNR)
>> Via Nomentana 56, 00161, Roma, Italy
>> Tel: +390644161535
>> Fax: +390644161513
>> <mailto:aldo.gangemi@istc.cnr.it>aldo.gangemi@istc.cnr.it
>>
>> <http://www.loa-cnr.it/gangemi.html>http://www.loa-cnr.it/ 
>> gangemi.html
>>
>> icq# 108370336
>>
>> skype aldogangemi
>
>
> --
> ---------------------------------------------------------------------
> IHMC		(850)434 8903 or (650)494 3973   home
> 40 South Alcaniz St.	(850)202 4416   office
> Pensacola			(850)202 4440   fax
> FL 32502			(850)291 0667    cell
> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>
>



_____________________________________

Aldo Gangemi

Senior Researcher
Laboratory for Applied Ontology
Institute for Cognitive Sciences and Technology
National Research Council (ISTC-CNR)
Via Nomentana 56, 00161, Roma, Italy
Tel: +390644161535
Fax: +390644161513
aldo.gangemi@istc.cnr.it

http://www.loa-cnr.it/gangemi.html

icq# 108370336

skype aldogangemi
Received on Thursday, 21 June 2007 22:24:55 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:00:48 GMT