0

While making good ontology is for sure big and mostly manual effort, it could be interesting to know if there are any techniques / tools, which automate making vendor-specific, intermediate ontologies for ETL process, given, say, rich-enough JSON examples combined with API documentation.

I am aware of Linked Data standards and techniques, but may be there is something in the form of library to make a draft RDFS+ ontology from API call responses?

For example, there are some libraries to guess JSON Schema (I have even written a primitive one myself), so the task does not seem to have problems in theory.

Please, do not consider this question as "software recommendation" one, because I doubt there is any software for this, but at least educated guess at direction I can take with this. I also believe, this is very important for semantic / linked data projects, and I wonder why I can't find any hints and need to ask for help from more experienced here.

Roman Susi
  • 4,135
  • 2
  • 32
  • 47
  • Life would be made easier for you if the API provides JSON-LD (http://json-ld.org/) the @context annotation where the ontology is referred to in the graph could then be parsed. If your looking for some super tool that can convert any JSON structure into an RDF graph then that's going to be troublesome. Realistically you would need a specific transform implementation for a specific API. – Anthony Hughes Nov 23 '15 at 13:46
  • The idea came to me after doing specific ETL transforms three times. Maybe I had too good APIs to deal with (at least two of them), but transforms were more or less mechanical, and following roughly the same steps. That is why I thought that it is possible to semi-automate drafting an ontology. So, actually, the problem is on par with turning relational db to RDF (and have RDFS-something as bonus), no? The draft could be refined in other tools, of course. (actually, if such a tool would be available, it could accelerate SemWeb adoption a lot...) – Roman Susi Nov 23 '15 at 14:20

0 Answers0