Classification and Specialization

Classification is an activity to relate an individual thing to a suitable kind of thing (a concept or a class or a type), such that the kind of thing characterizes the nature of the individual thing. The result of a classification activity is a classification relation between the individual thing and the kind of thing. Such a classification relation (a relation type) is denoted in Gellish by the standard phrase <is classified as a>. That relation type has Unique Identifier (UID) 1225.

Examples of classification relations are:

New York <is classified as a> city
M1 <is classified as a> motorway
L1 <is classified as a> length
A1 <is classified as> cleaning
John <is classified as a> man
D1 <is classified as a> inspection report
R1 <is classified as a> composition relation
John <is classified by role as a> teacher

The above examples illustrate that not only physical objects can be classified, but any individual thing, which includes also things such as products, properties, activities, processes, people, documents, units of measure, mathematical and geometric concepts, roles and even relations.

Most classification relations refer to the nature of a thing, the last example classifies John as a teacher. However, teacher is not a nature of a person, but it is an example of a (temporary) role of a person. Therefore, that is an example of a classification by role. Because of the difference in meaning between classification by nature and classification by role, the relation type that is used is different from an ordinary classification.

The definitions of the kinds of things are provided in a base Formal English Dictionary-Taxonomy. It consists of a collection of Domain Dictionaries-Taxonomies that enable the classification of individual things by Formal English concepts in a large number of domains, including also relations. Together with the standard syntax (the Gellish Expression Table) this enables computerized storage and interpretation of Formal English expressions.
As the dictionary-taxonomy is extensible with proprietary concept definitions and relations, the language becomes open and generally applicable.

Benefits of proper classification

Proper classification and concept and terminology management in organization is very practical and powerful for the following reasons:
Knowledge, requirements and documents is by definition related to the kinds of things (concepts, classes) about which they provide information. When the kinds of things and their names (terms) are included in a formal dictionary-taxonomy and when the information, knowledge and requirements are related to those concepts, then the combination becomes a knowledge base in which the kinds of things and their relations become a powerful navigation mechanisms towards definitions, knowledge and requirements about such kinds of things. Furthermore, when individual things are properly classified, then software can inform the users about information, knowledge and requirements that are applicable for those individual things.

For example, assume a dictionary-taxonomy contains a definition of the concept 'motorway' and a knowledge base contains the specification of a requirement R1 about roads and R2 specifically about motorways and a database contains the statement that M1 is a motorway. The three expressions will be as follows:

motorway <is a specialization of> road
R1 <is a requirement about a> road
R2 <is a requirement about a> motorway
M1 <is classified as a> motorway

Then software can be used to find requirements about motorways, which will result in R1 and R2, because of the definition that a motorway is a specialization of road. The software can also inform a user that R1 and R2 are applicable to M1.

Thus, by classification of individual things by concepts from a dictionary-taxonomy, it becomes explicit which definition applies and which knowledge and which requirements are applicable for all those the individual things. This makes communication less ambiguous. It also enables to apply the knowledge and requirements during design processes and its enables to use the knowledge and requirements during verification processes.

The definition of more specialized concepts, has the advantage that it becomes possible to be more specific about the allocation of knowledge and requirements that are only applicable for specific specialized subtypes. If that knowledge and requirements would be allocated to more generalized kinds of things, then it seems as if they are applicable for all the subtypes. This would cause that users are confronted with knowledge and requirements that are not applicable for their objects. The use of specialized subtypes enables to be more specific to classify things to specialized kinds and thus be selective about what is applicable and what not. Sometimes people object that this will result in too many subtypes. However, the subtype-supertype hierarchy and the inheritance mechanism enable software to hide the subtypes where necessary, so that the users are only confronted with the subtypes they need.

Classification versus Specialization

Classification relations should be distinguished from specialization relations. The first ones are relations between an individual thing and a kind of thing, whereas the latter are relations between a subtype kind and a supertype kind. A specialization relation is defined as a relation between two kinds that specifies that the subtype has all the aspects that the supertype has, whereas one or more aspects of the subtype are more constrained than that aspect of the supertype. A specialization relation is denoted in Gellish by the phrase <is a specialization of> with UID 1146. Examples of specialization relations are:

two lane road <is a specialization of> road
vacuum cleaning <is a specialization of> cleaning
man <is a specialization of> person
teacher <is a specialization of> role
customer <is a specialization of> role
red <is a qualification of> color (colour)
bar <is a qualification of> pressure scale

The above examples illustrate some relations that are used in the Gellish Dictionary-Taxonomy in various domains. The last two examples illustrate the use of a subtype of the specialization relation. Each of those qualification relations is a relation between a particular qualitative concept (red and bar) and a generic concept (colour and scale).

Classification systems

Sometimes the term classification is used to refer to the development of a classification system, being a collection of kinds of things (classes) arranged in a coherent structure, typically a hierarchy. The classes in such a classification system are intended to be used to classify individual things or to act as an entry point on a route to find knowledge (facts or documents). A Gellish Dictionary-Taxonomy is such a classification system. However, in contrast with many existing classification systems, a Gellish Dictionary-Taxonomy is a pure subtype-supertype hierarchy. A main advantage of a pure taxonomy is that the inheritance mechanism can be used, which means that all knowledge and requirements about a kind of thing is also applicable (inherited to) all the subtypes of that kind. Many other classification systems appears to be a hierarchy that is a mixture between collection relations (groupings) and subtype-supertype relations. As a consequence the inheritance mechanism does not work for such classification systems. The Gellish Modeling Methodology provides guidelines for mappings between existing classification systems and the Gellish Taxonomy. The method maintains the definition of the collections of classes and a mapping to the classes in the pure subtype-supertype hierarchy. This makes integration and combined usage perfectly feasible.

Explicit versus implicit classification relations

In the Gellish modeling methodology all classification relations are explicitly represented in a database. In conventional databases the classification relations are usually implicit. (typically they are represented by an instantiation of an entity type or attribute type). This conventional implicit classification limits and fixes the variety of classes that are used and limits the inheritance of aspects and relations from supertype classes to the ones that are predefined. Thus it also limit the application of knowledge and requirements during design and verification processes. Such a constraint does not exist in Gellish, where the number of classes is flexible.

The Gellish Domain Dictionaries-Taxonomies provide an extensible subtype-supertype hierarchy of concepts that enables a detailed classification of any kind of thing.

Full article

The full article (for licensees only) provides further guidance on how to classify products, properties, processes, documents, etc. and on the classification process and the use of knowledge and requirements about kinds of things (classes), for example in Computer Aided Design as well as in Computer Aided Verification of designs and in procurement of products.