1397RE: [TaxoCoP] Visualization technologies
- Jul 5, 2006
Hi Bob: Good feedback.
Please consider the following responses.
I use hierarchy and “tree” differently, now referencing some knowledgeable content management system (CMS) folks at http://www.cmswiki.com/tiki-index.php?page=Classification. (Follow the Taxonomy link for their hierarchy vs. tree discussion.)
According to my understanding of their material (which also fits my own research so is comfortable for me), hierarchies have inheritable attributes, and my own “categorization” relation type is intended to be such a hierarchy, not only of “classes” but also of “instances” of those classes, which might suffice as your “instantial” relation.
I use this “categorization” relation type in building up online “reference catalogs” (i.e., plain-speak for taxonomies) of types and instances of the categorized things, which I subsequently use for “lookups” in online forms and such to provide consistency in adding this reference as broader context and narrower content for the form’s subject.
Your “partitive relation would probably correspond to my “containment” relation type, representing a “tree” structure (with non-inherited attributes) of different “types” of things (e.g., your escalator example). I typically describe my “containment” relation type as “container/component” or “distribution/composition” types of relations.
The remainder of my relation types (i.e., sequence, change, variance, equivalence, and descriptive) would also be “trees” of diverse things in space, energy, matter, time, or concept.
(By the way, when a user filling in a form creates a “relation instance” of these seven relation types between my seven root taxonomies (i.e., generalized subject categories or references catalogs) during the process of selecting a catalog’s “lookup” value, the user is actually defining a “semantic thread” (i.e., a compound assertion consisting of multiple triples/assertions) that, when taken in the aggregate for a particular user’s current viewpoint, forms an “ontology” of the user’s subject “world view”.)
As to TheBrain’s capabilities, I didn’t say it couldn’t create the various relation types, it is just weak in their use (i.e., only the “containment” network relation is “managed” by TheBrain’s UI.)
CommIT Enterprises, Inc.
Roy Roebuck III
Senior Enterprise Architect
+1 (703) 486-5540
+1 (703) 486-5506
From: TaxoCoP@yahoogroups.com [mailto:TaxoCoP@yahoogroups.com] On Behalf Of Bob Bater
Sent: Wednesday, July 05, 2006 5:54 PM
Subject: RE: [TaxoCoP] Visualization technologies
Thanks for your interesting comments, which I have excerpted below. It is good to hear that some have used the Brain since its first beta release (me too) and are still using it.
However, I obviously have some learning to do here, because I don't understand why a hierarchy must have inheritable attributes to be a hierarchy. I remember that this was discussed fairly recently in this group, but I don't remember the conclusion (if any).
The hierarchies (taxonomies, thesauri) I generally find myself building recognize just three types of hierarchical relationship: generic, partitive and instantial. The generic relationship, I guess, is equivalent to your 'categorization' and the partitive to your 'containment', but I'm not sure which of your relations the instantial relationship equates to. Nor do I see how the containment relation type can always support inheritance. An escalator for instance, contains the major components 'steps', 'handrail', 'electric motor' etc.. But since the steps are usually metal and the handrail rubber or plastic, I can't see what's being inherited here, except membership of the top-level class (although of course, 'handrail' could also be a member of an entirely different class). But perhaps you wouldn't regard this as an example of a taxonomy?
I am also interested to know why you say The Brain has "only limited support for categorization, sequence, change, variance, equivalence, and descriptive relations" when it allows you to assign a link type to a link between two terms or concepts and to show those links in different colours. I regard this as equivalent to Topic Maps allowing the typing of associations. But again, perhaps there is more to representing these other types of relation than just naming them?
>>Principal Associate<<>>>> Bristol , UK <<<<
On 05 July 2006 18:22 Roy Roebuck said:
... it is less capable at creating “hierarchies” (i.e., classification structure with inheritable attributes) of content, which are fundamental to taxonomies and ontologies.
I typically categorize relationships (i.e., verb phrases) into one of seven “Relation Types”: categorization, containment, sequence, change, variance, equivalence, and descriptive. TheBrain has strong support for containment relations, but only limited support for categorization, sequence, change, variance, equivalence, and descriptive relations. Other technologies focus on one or more of these other relations.
- << Previous post in topic Next post in topic >>