I am working on a semantic application that requires keeping track of user information for whenever they change certain data within the individuals of the model.
I also need to have some auth functionality for the actual web application.
Should the auth mechanism be done in a relational model? Or am I better off by baking in the auth modeling in the ontology so that I can avoid having a relational and a semantic model?
How are such things typically handled?
asked 04 Apr '11, 18:53
Please have a look at the FOAF Vocabulary for basis user modelling and WebID for an authentication protocol that is part of the Semantic Web technology stack. Furthermore, Web Access Controlo, the Privacy Preference Ontology and the Cognitive Characteristics Ontology might be interesting in this context.
answered 04 Apr '11, 19:22
I think whether or not account data should be in a semantic model depends on your internal requirements. Of course it's pleasing to be consistent and just use one paradigm and one store for both tasks and you could also query across the data. But maybe you have to keep them separate for security, performance or other reasons. It also depends on whether you actually need to expose the user account data to the outside world.
For a university project I had a similar problem and I pragmatically used a semantic model for everything (even user passwords and things like whether their email addresses have been verified) but I stored them in such a way that the public and the private data were in separate stores (so you could never get to the passwords from a public endpoint) while still being able to query across both datasets internally.
answered 04 Apr '11, 20:04