You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Reverse engineering the current PyGraft implementation, notably looking at pygraft.kg_generator.InstanceGenerator.load_schema_info(), shows that no code change to PyGraft is needed to achieve the above-mentioned feature. For instance, translating an existing ontology implementation (e.g. in Turtle syntax) into the PyGraft class_info.json and relation_info.json data structures expected by PyGraft (see attached files) would be sufficient to benefit from the process inherent to the generate_kg task.
Hence, the quick win solution would be to implement some kind of UserDefinedSchema2PyGraftConfig side script.
Describe alternatives you've considered
Implementing a complementary InstanceGenerator class where the load_schema_info() method would directly take an ontology as input.
Additional context
It seems that the schema specifications, as per the current class_info.json and relation_info.json structures, could be sufficient for a first release of the UserDefinedSchema2PyGraftConfig script and the generate_kg task.
Is your feature request related to a problem? Please describe.
Discussing options to achieve the
Allow support for any input schema (and not only the schemas generated by PyGraft)
described in https://github.com/nicolas-hbt/pygraft?tab=readme-ov-file#upcoming-featuresDescribe the solution you'd like
Reverse engineering the current PyGraft implementation, notably looking at
pygraft.kg_generator.InstanceGenerator.load_schema_info()
, shows that no code change to PyGraft is needed to achieve the above-mentioned feature. For instance, translating an existing ontology implementation (e.g. in Turtle syntax) into the PyGraftclass_info.json
andrelation_info.json
data structures expected by PyGraft (see attached files) would be sufficient to benefit from the process inherent to thegenerate_kg
task.Hence, the quick win solution would be to implement some kind of UserDefinedSchema2PyGraftConfig side script.
Describe alternatives you've considered
Implementing a complementary
InstanceGenerator
class where theload_schema_info()
method would directly take an ontology as input.Additional context
It seems that the schema specifications, as per the current
class_info.json
andrelation_info.json
structures, could be sufficient for a first release of the UserDefinedSchema2PyGraftConfig script and thegenerate_kg
task.class_info.json
relation_info.json
The text was updated successfully, but these errors were encountered: