The Abstract Visual Model (AVM) is an intermediate RDF model that explicitly models graphic relations and roles of graphic objects. The concepts, such as Graphic object, Graphic relation and various syntactic graphic roles needed for modelling are taken from the graphic module of the Visualization Ontology (VISO).
An example of an AVM representing an Undirected Linking relation is given below. Two graphic objects are linked with each other via a third graphic object, playing the role of a connector:
The AVM may be written to a file for debugging purposes, see test
Note on Superroles
To consider also the superroles that graphic objects automatically play (e.g., each object playing the role of a "start node" also plays the role of a "node"), we have to consider triples from the VISO/graphic model. In the current implementation this is not solved by extendeding the AVM and doing RDFS reasoning (this works as well, but currently adds much noise to the AVM). Instead the superroles are queried while doing the a SPARQL query to access the roles. See also AVMUtils#getRolesAsURIForGO. Handling superroles allows for using CSS styles on them, for example to set the size of all nodes to some constant value, which is not the task of an RVL mapping (see Assigning Constants).