Validator for Multimodal Knowledge Graph Functional Dependencies

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 .

Usage

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

Generating Knowledge Graph Functional Dependencies

See https://gitlab.com/wxwilcke/mkgfd