Usage

The apply command takes a single KGCL instruction specified with the -k (or --kgcl) option, or a KGCL file (with one instruction per line) specified with the -K (or --kgcl-file) option. Both options can be used repeatedly, to specify more than one change (with -k) or changes from more than one file (with -K). It applies the change(s) to the current ontology loaded in ROBOT.

For example, the following command will apply the changes described in the changes.kgcl file to the ontology read from input.ofn, and save the modified ontology in output.ofn:

$ robot apply -i input.ofn -K changes.kgcl -o output.ofn

To apply a single change directly from the command line instead:

$ robot apply -i input.ofn -k "create class EX:999 'new class'" -o output.ofn

As for any other ROBOT command, the -i and -o options may be omitted if the apply command is not the first or last command of the pipeline:

$ robot merge -i input1.ofn -i input2.ofn \
        apply -K changes.kgcl \
        annotate --ontology-iri http://example.org/test -o output.ofn

It is also possible to create a brand new ontology by using the --create option instead of --input. In this case a new ontology will be created from the changes (obviously this only makes sense if the changes include at least some create instructions such as create class, create relation, etc.). The newly created ontology will become the current ontology of the ROBOT pipeline.

If some changes cannot be applied because they don’t match the contents of the ontology (for example, a change that tries to add a label to a class that does not exist), they will be written back to a file with the same name as the original KGCL file (the first KGCL file, if more than one has been specified), with a .rej extension appended. Use the --reject-file to specify the name of another file where to write the rejected changes, or -no-reject-file to disable writing back rejected changes.

The other changes (those that can be applied) will be applied normally, unless the --no-partial-apply option is used – in which case the command will refuse to apply any changes at all unless they can all be applied.

Some changes require the use of a reasoner to check if they can be applied. As for other ROBOT commands that need a reasoner, the reasoner to use can be specified with the --reasoner (or -r) option. The default reasoner is ELK.

If the --default-new-language option is specified, any NodeChange operation that does not have an explicit language tag will use the specified default new language tag.

Automatically assigned IDs for new entities

All KGCL operations require that you know the identifiers of the affected entities. When creating a new entity, this can be annoying as it means you need to figure out the identifier of the entity to be created.

To avoid this, you can get the apply command to automatically assign identifiers to newly created entities, by referencing them with pseudo-identifiers in the AUTOID: namespace. For example, you can do the following:

create AUTOID:1 "my new class"
add definition "the definition of my new class" to AUTOID:1
create edge AUTOID:1 rdfs:subClassOf EX:1234

to create a new class, give it a label and definition, and make it a subclass of the (supposedly existing) EX:1234 class, without having to know in advance what the identifier of the new class should be.

Automatic identifiers can be assigned using three different modes, described below.