Maps, cross-sections and profiles have subtypes that are diagrams and subtypes that are images. Thus, we cannot have a complete hierarchy for each one of them under solely diagram or image.
The possibilities to solve this problem are:
1) Split the three classes between diagram and image.
This solution would solve the problem, but wouldnt change the fact that experts group these entities without this distinction of image and diagram. Thus, the ontology wouldnt reflect the domain knowledge.
2) Move map, cross-section and profile directly under Figure.
This solution reflects exactly the understanding of the domain by experts. However, we would lose important information, which is the fact that these entities have subtypes that are mages (non-intentional) or diagrams (intentional).
3) Make map, cross-section and profile subtypes of both image and diagram.
In theory, this solution solves the problems presented in 1) and 2), but adds a multi inheritance problem. Although it is ontologically correct, this approach can lead to problems in future integrations of the ontology. This is also not recommended by many ontology development guides, such as protégé 101 and BFO book.
4) Move map, cross-section and profile directly under Figure AND create equivalent classes for diagram and image to use reasoning to infer the correct classification.
This is the approach we decided to follow. We have moved all the three to figure, but we created diagrammatic and image counterparts of these classes. The idea is that we add all the subtypes of maps (and the others) in the map class under figure, but, for each subtype, we add properties that allow us to infer with the support of a reasoner if it is a diagram or an image.
Maps, cross-sections and profiles have subtypes that are diagrams and subtypes that are images. Thus, we cannot have a complete hierarchy for each one of them under solely diagram or image.
The possibilities to solve this problem are: 1) Split the three classes between diagram and image. This solution would solve the problem, but wouldn
t change the fact that experts group these entities without this distinction of image and diagram. Thus, the ontology wouldnt reflect the domain knowledge.2) Move map, cross-section and profile directly under Figure. This solution reflects exactly the understanding of the domain by experts. However, we would lose important information, which is the fact that these entities have subtypes that are mages (non-intentional) or diagrams (intentional).
3) Make map, cross-section and profile subtypes of both image and diagram. In theory, this solution solves the problems presented in 1) and 2), but adds a multi inheritance problem. Although it is ontologically correct, this approach can lead to problems in future integrations of the ontology. This is also not recommended by many ontology development guides, such as protégé 101 and BFO book.
4) Move map, cross-section and profile directly under Figure AND create equivalent classes for diagram and image to use reasoning to infer the correct classification. This is the approach we decided to follow. We have moved all the three to figure, but we created diagrammatic and image counterparts of these classes. The idea is that we add all the subtypes of maps (and the others) in the map class under figure, but, for each subtype, we add properties that allow us to infer with the support of a reasoner if it is a diagram or an image.