Skip to content
Wouter Beek edited this page Dec 30, 2015 · 1 revision

1. Internal quadruple representation

Observations:

  • Named graphs (according to the data) are currently conflated with the ability to load data into different graphs (according to the application).
  • Originally, Line was the line in the file from which the triple was loaded. library(rdf_persistency) uses Line as time stamp to correlate transactions on different graphs. Caching applications might use it for last access, etc. Possibly we should separate that.

Proposal:

  • What about saying that a triple is identified by and have a predicate rdf_triple_property(S,P,O,G,Property) to query additional properties (last access, line, etc.)?
  • In the internal storage format for statements the 4th argument must be the graph name (either an IRI for named graphs or the default graph). We need a new rdf/5 (probably with a different name to query).

2. Default graph name

The default graph is currently called user.

Proposal:

  • Renamed the default graph from user to default.

3. Graph statistics

RDF statistics does not show the correct number of graphs in case a graph is unloaded:

?- use_module(library(semweb/rdf_db)).
true.
?- rdf_create_graph(test).
true.
?- rdf_unload_graph(test).
true.
?- rdf_statistics(graphs(N)).
N = 1.
?- rdf_graph(G).
false.
Clone this wiki locally