Validate any (RDF) Knowledge Graph against a set of Multimodal Knowledge Graph Functional Dependencies (MKGFD). Outputs a set of reification triples which flag candidate violations:
_:<bnode> a rdf.Statement;
rdf.subject <subject>;
rdf.predicate <predicate>;
rdf.object <object>;
rdfs.comment "Constraint Violation (p = <probability>) on: <constraint>"@en .
run_mp.py [-h] -c CONSTRAINTS [-d MAX_DEPTH] -i INPUT [INPUT ...]
[-n NPROC] [-p MIN_PROB] [--test]
optional arguments:
-h, --help show this help message and exit
-c CONSTRAINTS, --constraints CONSTRAINTS
Constraints to validate against (.pkl)
-d MAX_DEPTH, --max_depth MAX_DEPTH
Maximum depth to validate
-i INPUT [INPUT ...], --input INPUT [INPUT ...]
One or more RDF-encoded graphs
-n NPROC, --nproc NPROC
Number of cores to utilize
-p MIN_PROB, --min_prob MIN_PROB
Minimum probability to consider
--test Dry run without saving results