Network structureThe semantic network as constructed is composed of a few elements:
- Nodes or concepts
- Statements linking two nodes in a meaningful way
- Attributes, also called properties, of nodes, statements or other attributes
NodesNodes are perhaps the least interesting things in the network.
Why this statement, you might ask.If you look in detail to nodes you will discover that it is nothing more than a collection of attributes and a second collection of start/end points of statements. If we communicate about such a concept (node) we transfer part of the information we know about it: e.g. one of the names.
Statements have their meaning defined by a predicate and form triples: <subject><predicate><object> like in RDF. These three elements are all nodes in the network.
The predicate used in statements has in most cases, but doesn't need to, an inverse predicate associated that is used for the inverse relationship. The reason why not one bidirectional relationship with different labels is used, is that it allows relationships in opposite directions to have different attributes.
Attributes are defined by an attribute type (providing the meaning of the attribute) and a value. The attribute type is a node in the network, the value on the other hand is not. Attributes can be used for real life properties like a birth date, but also occur for many technical information purposes: internal and external id's, creation date, owner, rights etcetera.