[Top] [All Lists]

Re: [ontolog-forum] Endurantism and Perdurantism - Re: Some Comments on

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: William Frank <williamf.frank@xxxxxxxxx>
Date: Sat, 21 Mar 2015 14:47:21 -0400
Message-id: <CALuUwtBXyq6g3=BwODduTgYieWPKBoPZtg2ba3GGmB9roK4irQ@xxxxxxxxxxxxxx>

On Sat, Mar 21, 2015 at 1:22 PM, <rrovetto@xxxxxxxxxxx> wrote:
Important points.
1) Questioning the hierarchical structure of categories and ontologies

Right, I have found it counterproductive to create deep hierachies of conceptual categories.  Either of apriori, metaphysical, upper ontology categories, on the one hand, or domain specific concepts, on the other, when these concept are about human artifact, like securities, wine, or motor vehicles.  I have found that only when the categories are organizations used in natural science, especially biology, are deep hierachies useful.

Instead, I find that when people create artifacts,that is, talking from a domain specific point of view, they design them from features, or some such, each of which will have its OWN relatively shallow categories, for wine:  kinds of GRAPES, gegraphical locations, sugar content ranges, color, ...   So, when we create a new wine, we don't say, where does it fit in the hierarchy of wines, but which of *each* of these features do we want to apply to it.   Especially, as an off list remark from a data model person said  "when the rocket scientists design a new security, they do so by deciding where it fits in the hierarchy of securities - NOT!"  They instead compose it from the parts that will give it the desired characteritics, maturity terms, recourse terms, etc. etc.    O-O design might apply a 'decorator' pattern to describe this, not try to squeeze this into a vastly multiply 'inheriting' 'tangle'.  Not that you can't if you want, just that it is something that has an easier solution using parts and composition.  A boat with wheels you can drive on the road might not best have its own section is a boat catalog.  It might most likely be in the runnabout section, as it is unlikely to have any instances among the mega-yachts. 

2) Questioning the nature of top-level categories and their relation to lower-level ones

Right, , talking from a conceptual category point of view -- how do we coalesce how we talk about the world, think about the world, percieve the world, and thus are able to make *any* true a-posteriori statements, then we have a different set of shallow hierarchy of categories of thought (and so simultaneously, categories of the reality that people can communicate about to each other and to themselves as they think).    

I find it useful to say that 'ripening' is a biological process, and a banana is a physical object.     But I do not find it useful to organize all the physical objects and all the processes in the world or even in biology together.  I imagine that a treatise on biology would naturally  assume we all understood that ripening was a process, bananas objects, and cellulose a characterization of a substance.  And not have to tell you such underlying a-priori facts as objects are made up of substances, as lakes are made up of water, and that objects have weight, but that processes and the substance, as a whole does not, (all the water in the universe' being something different from 'water').  

A dictionary tells you the grammatical category of a word, which is a rough proxy for the semantic category of the concept, and also tells you the domain in which a given meaning of the word applies, banana (noun), (in botany, any variety of the species bananum).  When I create an English sentence, I can't use a verb as the subject.  So too, if I am to create a useful domain ontology, I need an upper ontology to provide the a-priori constraints on the categories of thought I will countenace.  But we can usually re-cast concepts and parts of speech from one category to another. 

William, are you saying...
1) Concepts (what you call Conceptual Categories) are not inherently structured in a hierarchy?

I think that atomic concepts can be so structured, but the hierarchies are generally shallow.   And, when you create a concept such as a 'Sushi Bond', it is not understood to be a Sushi Bond because of its place in a hierarchy, but rather because of a long list of composed features.   And, that there are two fundamentally different kinds of concepts, metaphysical, logical, ontological categories, such as the kinds that Aristotle talks about in Categories, that we have to use to be able to think and know AT ALL, and the kind that express what we have observed, using those metaphysical cocepts, such as the observation that a shark is not a fish.

2) Conceptual Categories are those that ...
(a) are more general than biological and other scientific categories,
(b) are domain-neutral categories, and as such are
(c) categories from top-level (domain-neutral) ontologies, e.g. from dolce, bfo, gfo, misc.
(d) are created, not discovered
no.   They are not discovered in the SAME way as a-posteriori science discovers.     When I *discovered* that Indians of the Northwest said things like 'it is Williaming over there', I discovered that the way we organize inot conceptual categories what we observe is mutable, but at the same time got confirmation that people had built into their ability to think and observe some categories that were common, but emphasized differently in different languages.   They are certainly NOT created out of 'whole cloth'.  This is a deep subject, of course, and I am hoping that while we need a set of conceptual categories, that we don't all have to be philosophers, and start barking.  Thanks to this forum, I just read the excellent "The Necessity of Metaphysics", and it is this very kind of thing I hope *not* to have to take a position on.  I want to merely say that for practical reasons, having about 9 categories of being in an upper ontology is useful.  That they are not arbitrary, nor, at the same time, can be conclusively demonstrate that any given set is 'right'.  And, at the same time, that it is naive to think that conceptual categories are just as much out there in the world, independently from our thought, as is the difference between sharks and fish.

3) Biological Categories (let's generalize it to Scientific Categories) are those that...
(a) are posited by biology (some scientific discipline), e.g. Heart, or some species
right.  And not only sciences, but in a different way, engineering and professional disciplines like accounting, securities, automotives, wine, software, etc.
(b) are discovered
well, there is clearly an interplay between more fundamental paradigms, which are also closer to basic conceptual categories.  Is action at a distance possible?  This is not discovered, but based on accepting or not some more fundamental viewpoint.     The lines are fuzzy.  This is why philosophers have to argue forever about the difference between a-priori and a-posteriori knowledge, between analytic and synthentic knowledge, etc.   But rough and ready, the difference is pretty obvious.  
see the above
Would you (anyone) say this is correct...
4) The problem with many upper- or top-level ontologies is that they...
(a) implicitly assume that their categories are like Biological/Scientific Categories, i.e., they have mind-external referrents/correlates, etc.

No, it is that they implicitly assume that the domain ontology should simply be a **further extension** of the same classification scheme that **they* are using, instead of the domain ontology starting over with  subjects that organize the domain.  Instead, the upper ontology should be applied to the domain concepts as an aspect of each, not as a supertype of each.

(b) this can become dogmatic, an imposition, creating prohibitions, etc.

no, that this is not effective.   Like putting all the verbs at the end of the book.
(c) the top-level categories are more accurately correlated with linguistic or everyday mental conceptualizations

yes.   Though again, they also, of necessity, mut represent the actual fundamental organization of reality, in some way, shape, or form.  I do not want to take a position as to the **degree** to which these categories represent the *actual* fundamental organization of 'reality' versus just the way people **think** of reality.    I think that this would be an impossible thing to prove, though some people want to think about it enough to come to a rational decision for themselves, as did Tuoamas Thako, with whose views, especially as applies to logic, I agree with.  Only, I just hope we don't need to have a position to do ontological engineering.

What I mean to emphasize is that we can *recast" our thinking to look at things using different categories of thought, and when we do, we have to go through transfomation that are kinds of rules of inference, not 'axioms'.   For example, Newton's third law might be true or false.  The second order induction axiom for arithmetic is true of things isomorphic to the natural numbers, such as strings.  Those are good axioms.   Rules of inference are the foundations for thinking about ANY axioms.  

Thank you so much, Robert, for your taking an interest in my views on this matter.   I practice these views, and I find the practice effective.         

Robert Rovetto

On Sat, Mar 21, 2015 at 9:41 AM, William Frank <williamf.frank@xxxxxxxxx> wrote:
Here is another root of the problem with engineering ontologies:

The idea that our concepts are organized in HIERARCHIES.    This is false. The analogy with biological hierachies has been the pied piper of engineering ontology since its beginnings in 0-0 programming, ignoring the real advances in the area made in library science and in enterprise data modelling. 

Boilogical hierachies are ***not conceptual categories at all** the are discovered, extended, and modified as part of a science.   

And, when conflate this with conceptual categories, by starting with a 'upper ontology' that divides things into categories of being, that more or less map to the possible grammatical categories we find in multiple languages, you get a deep awkward hierarchy.   Awkward because these categories providing *logical, a priori constraints*, such as processes don't have weights, while physical objects do, not scientific principles. 
 with a grounding in domain knowledge.   (If you were taking a biology course, and were taught that processes don't have weights, and I had an inquiring, independent mind, wouldn't you wonder?)

For examples, Mark Twain said: 'I once tried to read a book in German, but I never found out what it was about, because when I got to the last chapter, it was missing, and that is where all the verbs were."  And, a nameless big mutual fund used the REA accounting model as an upper ontology to design their whole enterprise data model into three hierachies: one for assets, one for events, and one for parties, and then implemented it this way.  So, when they wanted to deal with selling (event) a customer (party) a bond (asset), they had to pull things together from very disparate conceptual organizations.

This is what happens:
On Fri, Mar 20, 2015 at 3:20 PM, Chris Mungall <cjmungall@xxxxxxx> wrote:
I have seen this kind of thing
with other upper level categories. For example, I've seen ontologies
where a 'disposition' hierarchy mirrors a process hierarchy. I can see
how it would be very possible to do the same kind of thing with an
object vs process distinction.

The categories of being **chosen** for an upper ontology are best used as **aspects** of the thingees (see Barabara Hall Partee) that provide high level semantic rules, which could, for example in UML, be represented as stereotypes, or any other way we want to distinguish the domain-specific classifications of thingees such as securities being a type of financial instrument, from the domain-indepedent underlying logic, such as whether the thingee is *being thought of* as something that happens or is *being thought of" as something that endures, or is *being thought of* as something that is a frame for describing other thingees, etc.

So this below, is absolutely right:

On Fri, Mar 20, 2015 at 3:20 PM, Chris Mungall <cjmungall@xxxxxxx> wrote:

"Different aspects are
being described in each hierarchy. You may dislike this separation
philosophically, but it serves a purpose. We're quite simple minded at
the end of the day, and prohibiting someone attaching a "mass" or
"width" property to a fruit ripening process appeals to us, and the
benefits of abandoning this aren't clear."

But the problem is, the 'axioms' (I would express them as syntactic constraints and rules of inference, and leave axioms to express domain knowledge), that tell us what you can say about processes such as drainging and instances of processes such as drainging Lake Superior, vs, what you can say about physical things  such as Lake Superior, vs. what you can say about a stuff such as water, and about sets and about parts and wholes, all have nothing in particular to do with the *pariticular* science of biology.  These general 'axioms' are logical, in the broader sense, and can be used in any domain, while scientific principles of biology are the special knowledge of the biologist. 

The fact that these have gotten conflated in ontology is evidence of the deep hole in which the work of engineering ontologists is done, away from the lights not only of actual software design and library science and data modelling, but that might also be shed by modern philosophy, with many analogs in science, where we have learned to stop talking about what something 'really' is, vs. how we are looking at it, and have learned to see scientific theories themselves not as things that are 'true' or 'false', but more or less effective and insightful, that can only approach that asymptote of what is really out there.


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

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

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    (01)

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