Wiki Termontography

Termontography is a multidisciplinary approach in which theories and methods for a multilingual terminological analysis of sociocognitive terminology theory are combined with methods and guidelines for ontology engineering. A clear distinction is made between conceptual modeling at a culture-independent level and a culture-specific analysis of units of understanding. Hence, the prototypical structuring of understanding is taken into consideration.

The newly coined term ‘termontography’ could be interpreted as the field composed of

1. Terminology: refers to the usage and study of terms, that is to say words and compound words generally used in specific contexts (WikiPedia).
2. Ontology: the study of the categories of things that exist or may exist in some domain (John F. Sowa: http://www.jfsowa.com/ontology/).
3. Terminography: compiling collections of the vocabulary of special languages (Pointer: http://www.computing.surrey.ac.uk/AI/pointer/report/section1.html).

Purposes of the termontography approach:

1. Describing ontologies with (multilingual) terminological information.
2. Structuring terminological resources with ontologies.

Termontography tools

• Categorization Framework Editor (CFE)

The CFE is a tool for the construction of a Categorization Framework (CF). A CF is a language independent concept model that can be used to classify terminological information, and that describes a certain domain by means of concepts and concept relations.

• Termontography Workbench (TW)

The TW is an application for the compilation of a (multilingual) domain specific corpus. This corpus may then be used to extract terminological information that can be linked to the CF. The output of this process will be a termontological resource, i.e. an ontologically structured terminological database, that can be used in specialized applications (for example the KBExplorer).

• Termontography Editor (TE)

The TE is an application to manage termontological resources. Terminological information, i.e. terms and term descriptions, can be added and classified using the CF. The CF can also be transformed into a concept model consisting of concepts and concept relations. These concepts and concept relations can then be linked to (multilingual) terminology.

Categorization Framework (CF)

A CF consists of the following items: term, category, meta category, term meaning, meaning, attribute, property, bi-directional relation and bi-directional relation instance.

• A term is a text string that is classified by one or more categories. Each term must be classified by only one language category.

For instance, the text string main classified by the French language category fr constitutes the French term main.

• A category belongs to a single superordinate concept we call a meta category and can be used to classify items, for instance terms.

The English language category en, for instance, belongs to the meta category language.

• A meta category specifies the superordinate concept of all the categories that belong to it.

If we add the category fr, for instance, to the meta category language, this category represents the French language.

• A term meaning is an item with both a reference to a term and a meaning, i.e. (meta) category. Term meanings may be added to a meta category and/or category.

With this definition a term meaning now refers to a certain meaning, while it still can be represented by a single term. The term meaning that references the English term joint and the category marijuana cigarette with meta category cigarette, for instance, clearly describes this term meaning as a marijuana cigarette (see figure 1).

• A meaning is the underlying item of a meta category and/or category. A meaning has a list of term meanings and may have references to both a meta category and a category.

The term meaning wind instrument, for instance, could reference a certain meaning with references to the meta category wind instrument and the category wind instrument. The meta category wind instrument then indicates that categories such as flute and trombone, which belong to this meta category, are wind instruments. The category wind instrument, however, could be used to classify something, for instance a bottle of beer, as a wind instrument. The meta category thus specifies the superordinate concept of a category, while the category is used to indicate a certain aspect of the classified CF item.

To further clarify the intended meaning of a concept relation (or concept), we allow properties to be added to CF items (categories and term meanings, for example). To implement properties, we introduce the items attribute and property.

• An attribute is implemented as a category with meta category attribute. Each attribute should refer to a certain value type. The list of possible value types depends upon the specific implementation of the CF. The value type text string is a minimum requirement. The value types URI , URN and URL have also proven to be very useful.

For example, the attribute description with value type text string could be used to describe a term meaning, while the attribute extra information with value type URL could be used to refer to a web page with extra information about a category.

• A property references an attribute and a value of a certain value type. Properties may be added to each CF item i.e. meta category, category, term, term meaning, attribute, property, bi-directional relation and bi-directional relation instance.

For example, we could add a property with attribute extra information and value http://en.wikipedia.org/wiki/Knee to the category knee.

To efficiently manage (meta) categories, relations are used. A relation between two CF items has two directions, mostly with a different meaning. Therefore, we introduce the item bi-directional relation to specify a bi-directional relation between two CF items.

• A bi-directional relation references at least one relation and at most two relations. Bi-directional relation instances can be created between meta categories, categories and term meanings.

A relation can be created by adding a category with meta category relation. We may add the categories is part of and has part, for example, to the meta category relation. A logical bi-directional relation should reference both these two opposing categories. Let us notate this bi-directional relation as ((is part of), (has part)).

• A bi-directional relation instance references a bi-directional relation and two meta categories, categories or term meanings. Since the direction of a bi-directional relation is usually relevant, the bi-directional relation instance makes a distinction between the source and the target item.

The bi-directional relation instance that references the bi-directional relation ((is part of), (has part)), the source category lower arm and the target category arm would indicate that a lower arm is part of an arm.

Figure 1: CF items (italic) in relation to their corresponding semiotic notions (bold).

---

Almost brilliant! To show true brilliance, try saying it again in little words my (very bright) six year old can understand.

Sometimes a cigar is just an Object.