I understood from your article that you use SHACL generated from the DDL which is then used to generate TARQL scripts. After there is more SHACL to convert the result into the wanted knowledge graph vocabulary.
I'm very curious if you tried sparql-anything and if that would not be a more suitable approach compared to TARQL?
I view sparql-anything and tarql both as being indispensible tools; SA is generally better when dealing with mixed heterogenous data, but Tarql has marginally better support for row-based systems.
I understood from your article that you use SHACL generated from the DDL which is then used to generate TARQL scripts. After there is more SHACL to convert the result into the wanted knowledge graph vocabulary.
I'm very curious if you tried sparql-anything and if that would not be a more suitable approach compared to TARQL?
I view sparql-anything and tarql both as being indispensible tools; SA is generally better when dealing with mixed heterogenous data, but Tarql has marginally better support for row-based systems.