Sem (Semantic Memory)
Data Structure
updated 2001.4.20

It is proposed to Sem-dev that this labeled sequenced arc

be represented internally to a sem application
in the following relational database structure...

  • A pointer can point into one of three spaces:
    • Nodes - code = 0
    • Literal - code = 1
    • Statements - code =2
  • The node and object pointers designate the space they point into in the fields sP and oP respectively. The verb pointer can only point to another node.
  • Context is defined as a collection of statements associated with a node.
  • A node that reifies a statement points to that statement from the node field (i.e. sP=2).
  • If seq is 0,
    • then all arcs with the same node and verb are considered a rdf:Alt or a rdf:Bag depending on the semantics of the verb.
  • If seq is a positive integer,
    • then arcs with the same node and verb are labeled with the sequence and will be considered a rdf:Seq.
  • It is conjectured that this structure will require only two internally defined identifier arc labels to be able to read and write RDF\XML and Semenglish and round-trip it.  
  • semUri 
    • - domain = any node
    • - range = a URI stored in the literal space
    • - needed to support rdf:about and rdf:resource
  • semTitle
    • - domain = any node
    • - range = any English string stored in the literal space 
    • - needed to support Semenglish's ability to use English words and phrases as identifiers of nodes
  • Any node can have either, both, or neither  a semUri, and a semTitle.