Home About Subscribe Search Member Area

Humanist Discussion Group

< Back to Volume 33

Humanist Archives: Nov. 20, 2019, 8:04 a.m. Humanist 33.422 - non-hierarchical concept ontologies

                  Humanist Discussion Group, Vol. 33, No. 422.
            Department of Digital Humanities, King's College London
                   Hosted by King's Digital Lab
                Submit to: humanist@dhhumanist.org

        Date: 2019-11-19 15:54:52+00:00
        From: Jan Christophe Meister 
        Subject: Re: [Humanist] 33.417: non-hierarchical concept ontologies

Dear Herbert and fellow ontologists (thanks to everyone who has responded!),

I think that facet classification does indeed constitute a valuable
option. As far as I can make out it is in fact a mix of an upper-level
hierarchical layer which is ordered tree-like and results in unique
nodes at the end of each branch, plus 'facets' that cut across these
branches horizontally and serve to characterize nodes by assigning them
property/value-pairs. The best example of a fact classification system
is Ranganathans 'colon classification' (see
- I owe this find Michael Sperberg-McQueen.

Ian's 'aspects' seem to be something conceptually related, but situated
rather on the phenomenological level of primary empirical observation of
individual entities. We encounter an object, and in addition to
registering that "there is an X" we also find X to be qualified in terms
of an 'aspect', e.g. "there is an X and it is green". Of course, once
you start to abstract from the individually encountered aspectual "is
green" to the concept of "color" you have the first building block for a
potential 'facet' under which you could then systematically describe all
X, Y, Zs etc. irrespective of whether their nouns, dwarfs, unicorns etc.

In our particular use case (i.e., text annotation in CATMA, which takes
the form of external stand-off markup modeled according to the Web
Annotation Data Model and is serialized to file using the recommended
JSON-LD format for Linked Data) we already have something like 'facets':
Any CATMA tag can be further specified by (a) following the hierarchical
branch/node structure which it is part of until you reach a more precise
sub-tag, then (b) assign that (sub-)tag to the relevant string or phrase
so that you now have a tag instance defined by category and stand-off,
and (c) then qualify that instance in terms of a property/value
declaration. In other words our 'X' at string position 'n to n+1' might
be tagged as, say, 'beverage' (tag) / 'health drink' (sub-tag) where
'color = green'  (property/value-pair).

Now for the complications. Number one: given a set of categories = tags
we might be able to anticipate useful properties and their value-ranges
upfront - but often enough the practice of (hermeneutic or semantic,
i.e. interpretive rather than declarative) text annotation makes us
realize that we are missing a tag, sub-tag, property AND that we've
missed out on a potential value.  We've dealt with that scenario by
allowing users to extend a CATMA tagset on the fly.

Two, not only 'health drinks' may have the property 'color', but so do
'animals', 'cars' etc.. And it may turn out that the use of a specific
property becomes so high-frequent that the user eventually decides that
it would be preferrable to elevate 'color' from 'property' to 'category
= tag' (not to mention 'greenness' = property value). So far we're not
allowing for this in CATMA - but it is a common feature request, and it
presents not just a technological, but indeed a fascinating conceptual
challenge. Let us assume we would implement this as well: we'd be
consciously messing with the hierarchy of our (explicit or implicit)
'ontology' in that we would allow for transposing its elements across
categorial and systematic layers. Of course, many phenomena in natural
languages do just that - think of metaphor or any other figurative
trope, so the dynamic as such is nothing new. It's just that we battle
to acknowledge it as an epistemic affordance. And that's basically the
backdrop to my question concerning a potential 'non-hierarchical' ontology.



Am 19.11.2019 um 10:03 schrieb Humanist:
>                    Humanist Discussion Group, Vol. 33, No. 417.
>              Department of Digital Humanities, King's College London
>                     Hosted by King's Digital Lab
>                         www.dhhumanist.org
>                  Submit to: humanist@dhhumanist.org
>          Date: 2019-11-19 00:09:29+00:00
>          From: Dr. Herbert Wender 
>          Subject: Re: [Humanist] 33.408: non-hierarchical concept ontologies
> Chris and Lian,
> Remembering what I learned in the late 1970s about discussions in information
> and library sciences, I'm wondering what would be the difference between
> forthcoming 'non-hierarchical ontologies' and well-known non-hierarchical
> techniques when librarians classify texts? About faceted classifications f.e.
> read just in the opening sentences of the German Wikipedia entry:
> "Eine Facettenklassifikation (auch analytisch-synthetische Klassifikation) ist
> ein Klassifikationssystem, bei dem die Objekte eines Wissensbereichs nicht in
> eine relativ unflexible Baumstruktur eingegliedert werden, wie es bei rein
> hierarchischen Systemen der Fall ist. Stattdessen erfolgt die Einordnung eines
> Objekts durch Zuordnung mehrerer voneinander unabhängiger Begriffe."
> Is there a more than terminological difference between 'aspect-oriented
> ontologies' and such a faceted classification?
> Willard,
> Do you know of studies in language history comparing such a process of
> pluralisation with the transition from plural forms (arts, beauties) to
> (in german 'Kollektivsingular': die Kunst, die Schönheit)?
> Best regards,
> Herbert
> -----Ursprüngliche Mitteilung-----
>             Humanist Discussion Group, Vol. 33, No. 408.
> [1] From: Iian Neill 
>      Subject: Re: [Humanist] 33.398: non-hierarchical concept
> ontologies? (78)
> [2] From: Willard McCarty 
>      Subject: ontology, ontologies & hierarchy (23)
> --[1]------------------------------------------------------------------------
> Date: 2019-11-14 12:13:31+00:00
> From: Iian Neill 
> Subject: Re: [Humanist] 33.398: non-hierarchical concept ontologies?
> Dear Chris,
> I don't have any papers to reference on this concept as yet, although I
> will be touching on it a forthcoming paper, but in our Codex project we are
> exploring the application of something I call "aspect-oriented" ontology to
> texts. The name is taken from the use of "aspects" in programming languages
> which support attributes applied to classes and methods. Aspects are used
> to implement cross-cutting (horizontal) concerns in software -- likewise,
> in an "aspect-oriented" ontology there are concepts which can be thought of
> as cutting across type classifications. For example, the aspect
> "Florentine" may not only apply to persons but to objects, cuisine, schools
> of art, etc.
> Best regards,
> Iian
> ...
> --[2]------------------------------------------------------------------------
> Date: 2019-11-14 08:29:25+00:00
> From: Willard McCarty 
> Subject: ontology, ontologies & hierarchy
> Chris Meister's question about non-hierarchical concept ontologies leads
> me to wonder whether the pluralisation of 'ontology' in the late 1940s
> by Quine (who was not unfamiliar with digital logic and computing),
> followed by the quiet adoption of the term in computer science much
> later, provides some insight. Specifically, might it be the case that by
> pluralising the term ontological hierarchy is undermined?
> I had occasion to look into the history of 'ontology' for a workshop at
> Cambridge in 2017, the outcome of which was published in HAU: Journal of
> Ethnographic Theory 9.1 (2017): 147-61. See esp. pp. 149-51.
> Should anyone know of material related to 'ontology' in CS or elsewhere
> that I did not catch, I'd be grateful to know about it.
> Yours,
> WM
> --
> Willard McCarty (www.mccarty.org.uk/),
> Professor emeritus, Department of Digital Humanities, King's College
> London; Editor, Interdisciplinary Science Reviews
> (www.tandfonline.com/loi/yisr20) and Humanist (www.dhhumanist.org)


Dr. Jan Christoph Meister
Universitätsprofessor für Digital Humanities
Schwerpunkt Deutsche Literatur und Textanalyse
Institut für Germanistik
Universität Hamburg
Überseering 35
22 297 Hamburg
+49  40 42838 2972
+49 172 40865 41

Unsubscribe at: http://dhhumanist.org/Restricted
List posts to: humanist@dhhumanist.org
List info and archives at at: http://dhhumanist.org
Listmember interface at: http://dhhumanist.org/Restricted/
Subscribe at: http://dhhumanist.org/membership_form.php

Editor: Willard McCarty (King's College London, U.K.; Western Sydney University, Australia)
Software designer: Malgosia Askanas (Mind-Crafts)

This site is maintained under a service level agreement by King's Digital Lab.