Thank you for this awesome library!!
If there is an ontology imported to the model, it should be quite easy to map the data in the polar-df according to the prefixes, classes and constraints in the ontology, reducing the manual work of adapting the stOTTR-template.
If you name the columns in the df according to the ontology, you shoud be able to infer the predicates (from domain and range) and the objects in the OTTR-triple mappings.