(see below for use via Docker)
To run these examples, first:
cd tests/data
All matches, as TSV:
rdfmatch -f tsv -i basic.ttl match
Also save triples
rdfmatch -f tsv -i basic.ttl -G matches.ttl match
As above, use skos rather than owl as vocab
rdfmatch -f tsv -i basic.ttl -G matches.ttl --predicate skos:exactMatch match
Contrain to classes with http://example.org/x/ prefix:
rdfmatch -p x -f tsv -i basic.ttl match
New matches:
rdfmatch -p x -f tsv -i basic.ttl new_match
Cache index in a temp folder:
rdfmatch -X tmp -p x -f tsv -i basic.ttl new_match
Learn match rule probabilities:
rdfmatch -p x -X tmp -T -f tsv -v -i tests/data/basic.ttl -i tests/data/equivs.ttl learn > probrules.pro
Note this requires a training set - owl:equivalentClass is used for this
Once you you created the probabilistic rules, these can be applied:
rdfmatch -f tsv -i tests/data/basic.ttl classify probrules.pro
Generate mappings in SSSOM format:
rdfmatch -w prolog/rdf_matcher/obo_weights.pl -T -i tests/data/basic.ttl match
The library index_util is used for caching. Optionally this can go to disk
Use ./run.sh instead of rdfmatch. E.g.
./run.sh -p x -f tsv -i tests/data/basic.ttl match
Note that files must be in current directory or below, run.sh automaps the current dir to the container.