Published Subject Indicators for Whole-Part Subtypes

PSI Metadata

Description Many formal languages honor mereology relations, which link every Part of an entity to its Whole. CTM starts with the very general and useful "Whole-Part" association (often abbreviated to W-P)

W-P has many subtypes. Each has unique semantics, but their basic pattern soon gets familar. To manage it without inventing a mass of ad hoc names, CTM adopts a whole-oriented naming strategy for its subtypes, illustrated below.

Each association subtype places unique constraints on the role players of its instances. CTM lets them may be phrased in clear natural language, then published at a PSI that application programmers can use as coding specs.

CTM recommends that constraints be declared in constraint occurrences of W-P typing topics, phrased in a constraint language that for each sub-type of whole can limit the legal types, cardinality, and characteristics of all role player topics.

However the constraints get defined, they are really what distinguish the many subtypes of W-P. Human readers may get intuitive help from subtype names, but the constraints on legal blends of the role players is really what guides CTM-based decision making.

PublisherLexikos Corporation
CreatorDan Corwin
Languagehttp://www.topicmaps.org/xtm/1.0/language.xtm#en
Version2006/07/08
StatusPre-release CTM 1.0 draft for comment
Date Published2004/08/19

Index Of Subjects

Whole-Parthttp://www.lexikos.com/psi/ctm/patterns/#1
*...Wholehttp://www.lexikos.com/psi/ctm/patterns/#11
*...Parthttp://www.lexikos.com/psi/ctm/patterns/#12
Collectionhttp://www.lexikos.com/psi/ctm/patterns/#2
*...Set-Memberhttp://www.lexikos.com/psi/ctm/patterns/#21
*...Map-Memberhttp://www.lexikos.com/psi/ctm/patterns/#22
*...List-Memberhttp://www.lexikos.com/psi/ctm/patterns/#23
Substancehttp://www.lexikos.com/psi/ctm/patterns/#3
*...Compound-Parthttp://www.lexikos.com/psi/ctm/patterns/#31
*...Element-Isotopehttp://www.lexikos.com/psi/ctm/patterns/#32
*...Mixture-Parthttp://www.lexikos.com/psi/ctm/patterns/#33
Aggregationhttp://www.lexikos.com/psi/ctm/patterns/#4
*...Team-Memberhttp://www.lexikos.com/psi/ctm/patterns/#41
*...Group-Factionhttp://www.lexikos.com/psi/ctm/patterns/#42
Structurehttp://www.lexikos.com/psi/ctm/patterns/#5
*...Molecule-Parthttp://www.lexikos.com/psi/ctm/patterns/#51
*...Object-Piecehttp://www.lexikos.com/psi/ctm/patterns/#52
*...Organism-Piecehttp://www.lexikos.com/psi/ctm/patterns/#53
Situationhttp://www.lexikos.com/psi/ctm/patterns/#6
*...Relation-topichttp://www.lexikos.com/psi/ctm/patterns/#61
*...Verb-complementhttp://www.lexikos.com/psi/ctm/patterns/#62
Descriptionhttp://www.lexikos.com/psi/ctm/patterns/#7
*...Equation-Parthttp://www.lexikos.com/psi/ctm/patterns/#71
*...Picture-Parthttp://www.lexikos.com/psi/ctm/patterns/#72
*...Paragraph-Parthttp://www.lexikos.com/psi/ctm/patterns/#73


Whole-Part

Published Subject Identifier: http://www.lexikos.com/psi/ctm/patterns/#1

CG literature defines the
Part as one type of Component. The core W-P association type uses it as one role type, and associates it to the role player of a whole role type. The latter is the central topic in any W-P graph: CTM defines semantically similar relations by incrementally modifying the structure above in sub-types, often defined by adopting these naming conventions:
  1. A uniquely named type constraining legal Whole players
  2. A suggestively named subtype of Part for the others
  3. A subtype of Whole-Part, named by hyphenating names for 1 and 2
Rational naming rules help chemists identify organic chemicals, but also expose to them the internal molecular structure of each compound. If W-P association subtypes are named by equally helpful rules, that makes their expected usage more clear, and suggests one of the main constraints on employing it it.

By following this naming convention, the PSIs below depict three open lists of W-P graph subtypes, each of which slightly varies and extends the semantics of the baseline association above.

These short listings seem to "cover" a fair fraction of the topics discussed in English dialogs. So if you need to model the Parts of something, consider wrapping it in one of these primitive associations, or in any custom subtype you invent.

Additional W-P models can be easily added to CTM 1.1, as user feedback warrents.


Collection

Published Subject Identifier: http://www.lexikos.com/psi/ctm/patterns/#2

In Conceptual Graphs, the plural form of an entity has a special {*} notation which adds semantics similar to a set - the first of the abstract W-P subtypes below: The remaining subtypes add constraints: Lists demand a global order; Maps insist their Members have unique names. Such definitional constraints are subtle, global, and hard to spec except in PSI documentation like this. But programmers read, and software can then use the PSI URL as the ID of a data type, so that seems adequate.

Substance

Published Subject Identifier: http://www.lexikos.com/psi/ctm/patterns/#3

The above W-P graphs represent mathematical abstractions. Those below involve similar notions, but might apply better to topics modeling physical entities. To model specific subtypes of Substance, Mixture, or Group, CTM would suggest using an application-specific subtype of the above, with a suitably renamed whole. Then for that "domain", the "range" constraints on each role player could be tailored as needed to formally define it in machine-readable ways.

Aggregation

Published Subject Identifier: http://www.lexikos.com/psi/ctm/patterns/#4

Such W-P subtypes may be so complex as to need multiple subtypes of Part, which the simple CTM naming rules cannot reflect. A team, for example, may subtype one Member into a Leader role, or for sports teams, may define complex roles like "First baseman" or "Halfback".

Structure

Published Subject Identifier: http://www.lexikos.com/psi/ctm/patterns/#5

Unlike the
Collection graphs, these W-P types impose constraints on the physical organization of their Parts. Such constraints could take the form of extrinsic relations which pairs of Parts should exhibit. A good constraint language should be able to express such constraints formally on application-specific subtypes of these basic models:

Situation

Published Subject Identifier: http://www.lexikos.com/psi/ctm/patterns/#6

Similar effects truly dominate these subtypes, where a conventional set of
Case Role subtypes has arisen for modeling the Parts. Semantics models of natural language often focus heavily on these Even in these, the same W-P modeling strategy perists, and so does the CTM premise that a Whole acquires meaning from constraints on each Part. In practical terms, each verb meaning thus has expected meanings for each of its complement.

Description

Published Subject Identifier: http://www.lexikos.com/psi/ctm/patterns/#7

Linguistic models are only one type, and people also deal in models which are mathmatical, pictorial, etc. An important feature of all these types is that they emerge from the head of some human creator who needs mention. One can build special associations between an author and a work product, but quick, casual models may also just extend the parts of the work to include its author, a required part constrained to be sentient. More detail, if required and availiable, can go into a full model of the Describing situation.