uom-ontology-std
[Top] [All Lists]

Re: [uom-ontology-std] Philosophy and Conceptual Choices, was: Re: A me

To: martin.hepp@xxxxxxxxxxxxxxxxxxx
Cc: uom-ontology-std <uom-ontology-std@xxxxxxxxxxxxxxxx>
From: Pat Hayes <phayes@xxxxxxx>
Date: Tue, 22 Sep 2009 18:05:23 -0500
Message-id: <04B33DC4-82F5-48AC-AA80-246DAD945ABB@xxxxxxx>

On Sep 22, 2009, at 3:13 PM, Martin Hepp (UniBW) wrote:    (01)

> Dear Pat:
>
> John F. Sowa wrote:
>> PH> There is no evidence for, and considerable pragmatic evidence
>> > against, the thesis that ontology engineering is improved by
>> > approaching it with the tools of philosophy, and certainly not
>> > with the methodologies of contemporary professional philosophy.
>>
>>
> I think that is too bold a statement, and I would like to see the  
> "pragmatic evidence against the thesis that
> ontology engineering is improved by approaching it with the tools of  
> philosophy".    (02)

Oh, for a start: the fact that OBO is founded on the idea of a  
'continuant', a close-to-incoherent idea with its roots in 19th  
century Polish phenomenology.    (03)

>
> First: I hope you agree that ontologies are most effective when they  
> define such categories (e.g. classes or relationship types)
> for which the category membership of a given phenomenon
>
> 1. is consensual among a large amount of individuals,
> 2. is valid across multiple contexts, and
> 3. remains valid over time.    (04)

Certainly 3. And 1., provided that 'large amount' is not understood to  
mean anything like 'majority'. But 2., no. I think most useful  
ontologies are in fact precisely useful when they identify and  
delineate a particular 'context', such as bioinformatics. But perhaps  
this is not what you mean by 'context': this word, alas, is so  
flexible in its meaning as to be almost useless in discussions like  
this.    (05)

>
> (Also, we must find a good trade-off between granularity and ease of  
> population - too subtle distinctions put a brake on populating  
> respective knowledge bases, too coarse distinctions limit the degree  
> of automation for processing the information - but that point is not  
> needed for my argument.)
>
> This challenge is independent of the expressiveness of the formalism  
> and of the amount of axioms that we use for constraining the  
> intension of a conceptual element in the ontology.
> If the categories defined are weak in one of the three respects,  
> even richly axiomatized ontologies are of limited value.
>
> For example, ontologies that distinguish
> - between book copies vs. book titles,
> - between product models and products, or
> - between human beings and their roles (like "student")
>
> are more useful than those that lack those distinctions because the  
> former don't mess up apples and oranges for tasks that need to keep  
> them apart.    (06)

True, but this begs the question. Perhaps some tasks do not need them  
to be kept apart: and in that case, unnecessary distinctions can be  
actively harmful. But I would add, all this is beside the point of my  
original comment about philosophy.    (07)

>
> Now, from my experience in designing real-world ontologies and  
> teaching conceptual modeling to students, the latter now for almost  
> a decade now,
> finding such "good" categories and evaluating their fitness is,  
> *the* key challenge in ontology engineering.
> Everything else builds on top of that.    (08)

I agree, provided that 'good' is understood properly as relative to a  
context of use, probably a community of users. But real philosophical  
issues rarely, if ever, intrude into such communities' business. How  
many engineers or scientists take Hume's problem of induction  
seriously? But this is, still, the central philosophical issue that  
motivates most philosophical debate which is relevant to these  
disciplines.    (09)

>
> I have no formal education in philosophy, unfortunately, but Welty's  
> and Guarino's "OntoClean" work, and other contributions that employ  
> subtle distinctions of categories of existence that are mainly  
> rooted in philosophy, are the only significant guidance for that  
> challenge I am aware of. I can exactly confirm what Alan Rector  
> reported on OntoClean in 2002 - that it simplifies the argument for  
> better ontological choices.    (010)

OntoClean enforces some elementary distinctions and enforces a certain  
internal consistency regarding the description of changes in  
properties and relations. Certainly this is of use, just as strong  
typing is of use in designing large software systems, and for much the  
same reasons. But this is more like logical coherence than actual  
philosophy. Do not confuse philosophy with something like 'clear  
thinking'. Philosophy does require clarity of thought, but so do  
theoretical physics and political history.    (011)

>
> In 1911, Frederick Winslow Taylor in his work on "Scientific  
> Management" has given quite some evidence that finding the ideal  
> conceptualization of a problem or task benefits from scientific  
> methods in the course of the analysis, more than from pure  
> motivation or practical experience in the domain.    (012)

If true, that is strong evidence in favor of my thesis here.  
Scientific method is about as far as one can get from the practice of  
academic philosophy.    (013)

>
> I am deeply convinced that we should transfer his spirit to the case  
> of ontology engineering: that we need methodological guidance on how  
> to define the most efficient categories. Towards that end, 2000  
> years of experience in developing subtle distinctions among  
> categories of existence may be the only guidance that we have. I  
> would not dispose that asset lightheartedly.    (014)

Nor I. I dispose of it quite solemnly and carefully, after half a  
career spent studying and teaching it, and trying to see how it could  
possibly be relevant to anything of practical use to anyone, and  
trying to formalize it and discovering that the actual act of  
formalization typically boils off all the philosophical debate to  
vacuity, in the exact sense that whatever the outcome - if philosophy  
could ever establish such a thing - of any of the interminable  
debates, they have exactly zero effect upon the actual formalization  
itself. And watching philosophers trying to say something useful about  
actual problems in AI and cognitive science (such as the notorious  
'frame problem') and failing spectacularly to even understand the  
basic issues, because they can only perceive them as philosophical  
problems rather than essentially problems in cognitive engineering.    (015)

Pat    (016)

>
> Of course one can overdo it in the strive for a single, "objective"  
> category system - such is obviously practically impossible to  
> engineer under resource constraints and in a world with non-zero  
> degree of dynamics and subjective judgment.
>
> Best wishes
>
> Martin
>
> -- 
> --------------------------------------------------------------
> martin hepp
> e-business & web science research group
> universitaet der bundeswehr muenchen
>
> e-mail:  mhepp@xxxxxxxxxxxx
> 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!
> =================================================================
>
> Webcast:
> http://www.heppnetz.de/projects/goodrelations/webcast/
>
> Recipe for Yahoo SearchMonkey:
> http://tr.im/rAbN
>
> Talk at the Semantic Technology Conference 2009: "Semantic Web-based  
> E-Commerce: The GoodRelations Ontology"
> http://tinyurl.com/semtech-hepp
>
> Talk at
> Overview article on Semantic Universe:
> http://tinyurl.com/goodrelations-universe
>
> Project page:
> http://purl.org/goodrelations/
>
> Resources for developers:
> http://www.ebusiness-unibw.org/wiki/GoodRelations
>
> Tutorial materials:
> CEC'09 2009 Tutorial: The Web of Data for E-Commerce: A Hands-on  
> Introduction to the GoodRelations Ontology, RDFa, and Yahoo!  
> SearchMonkey http://tr.im/grcec09
>
> <martin_hepp.vcf>    (017)

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes    (018)






_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/uom-ontology-std/  
Subscribe: mailto:uom-ontology-std-join@xxxxxxxxxxxxxxxx 
Config/Unsubscribe: http://ontolog.cim3.net/mailman/listinfo/uom-ontology-std/  
Shared Files: http://ontolog.cim3.net/file/work/UoM/  
Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?UoM_Ontology_Standard    (019)

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