Aim. Characteristic for the construction industry is that many parties work separately on a project, collecting for themselves the information they need, and building their own model of the project. Some are involved in the whole project, others only in a part of it, some serve only a single aspect, others more. There is not a single model that includes the whole project, instead the project is spread along several aspect models. Data need to be exchanged between these aspect models, which demands for a good information infrastructure. This is offered - amongst others - by a collection of concepts and the relationships between these.
Concepts are generalizations of the carriers of information, for the construction industry: construction notions. Construction notions have been identified and defined in many documents, however, between those identifications and definitions exists a lot of noise. For an adequate information exchange it is of great importance to use notions that are free of noise as much as possible. And, with regard to the intensive use of information technology, the notions should be interpretable by computer applications. Ontologies, as meant in Web Ontology Language (OWL) offer a platform for this. OWL, however, only offers the tools and techniques, but it is the content that counts. Content needs knowledge and insight for identifying and defining the right notions and the relevant relationships.
Construction notions can be collected from existing classification systems, standards, regulations, catalogues, etc. These are the objects that carry the information and the kind of data about these objects that are collected and exchanged. For the exchange of data definitions of datatypes are essential, because it is just there that errors occur caused by misinterpretation. Datatypes are dependent of the context in which they are used, objects form these contexts. A datatype like 'height' is as such unambiguous, but can only be quantified with a dimension added to it (for example 'millimetre') leading to the correct interpretation of the assigned value, hence 'height' has to be extended with the dimension to be exchangeable: 'height (mm)', 'height (in)', 'height (km)'. Associated with the object 'Door', 'height' needs further specification, such as 'clear height', 'fitting height', etc. Generally these different height types can be grouped together by reference to geometry, but essential for Door as a whole is the height of the passage, which is directly related to the function of the Door. For the exchange of data about Doors, the passage height in the realm of the 'Passage' function is a determinative datum, this can be modeled by associating a 'Passage' function to the object 'Door', and associating a datum of the type 'height' with an associated dimension (e.g. 'mm') to that function. In Construction Concepts 'Door' is a subtype of, amongst others, 'Passage', 'Passage' has an element 'Opening' and 'Opening' has an associated 'Geometry' as well as a 'Location'. In this model the 'passage height' is included in the geometry of the opening.
Hierarchy. A Hierarchical grouping of concepts offers the possibility of several levels of abstraction. These have to be chosen carefully, because the properties of a supertype are inherited by its subtypes. In OWL a supertype is a selection of concepts with similar properties, a concept can be a 'member' of several supertypes, each with its own set of properties. Construction Concepts uses supertypes for different data domains, such as 'System' and 'Function', each with their own specializations of those domains. Concepts derive their properties from these domains. For example a 'System' is characterized by a 'Function', subtypes derived from 'System' have a specific function derived from 'Function'. For example a 'Building structure' has a 'Structural function'. Associated with 'Functions are 'Performances', such as 'Fire resistance' as a 'Performance' of an 'Occlusion function', which is a 'Function' of an 'Occlusion structure' having 'Door' as a subtype, which makes 'Fire resistance' a 'Performance' of a 'Door'.
Naming. Concepts are indicated by their name, but the name of a concept is ambiguous: names are language and orthography dependent, concepts can have more than one name or a name can point to different concepts. OWL offers the possibility to make concepts independent from names for identification by using a name-independent URI, names are then associated with the concept by means of an annotation. Construction Concepts uses this possibility, with as one of the advantages that the same model can be easily represented in different languages (currently English and Dutch), and in addition synonyms and homonyms are not problematic.