The knowledge graph is made up of two files: ontology.csv and ontology_labels.txt (also a CSV file, apologetically). These files are designed to be ready by the ontology.py script. The ontology.csv defines the (bi-directional) links between writers and attributes using the global ID. Each writer and attribute is described in the ontology_labels.txt file with their NAME or IAM Data Set Writer ID.
The columns of ontology_labels.txt are: global id, IAM id or attribute name, node type (writer-id or writer-attribute), and ID specific to the type.
For example: '107,022,writer-id,71': (1) 107 this is global ID. (2) This node is writer ID 71 (3) This writer is associated with IAM writer 022.
Note: The writer-ids are incremental from 0 to align to one-hot prediction vectors.
Writer profiles are also described in writer_profile.csv.