On Sun, Oct 26, 2008 at 5:35 AM, bryan rasmussen <rasmussen.br...@gmail.com> wrote: > So, If I have a database already established without necessarily a lot > of documentation about that database and you want to add some new > structure to the RDF, what process would be the best practice for > doing that? > > I'm thinking sort of a step 1, step 2, step 3 type methodology, for example > "well step 1 I like to run this Sparql query here that dumps all > unique RDF structures so I know what I'm dealing with" (note: suspect > it is probably impossible to have such a query but figured I would > give an example) > > step 2, I use the following tools to automate updating all RDF already > in the database with the new data where applicable > > step 3, ah there isn't any.
In my project with Virtuoso, what I did was have a seperation between the "schema" and the "data". Since I evolved this later and my schema was mixed up with the data, I was not in a position to use a seperate graph for the schema and another for the data. But, since all my schema subjects had a URI with a specific prefix, all I had to do is drop all triples which have any resource at subject position that had a URI matching this prefix. This let me evolve the schema easily .. however, it wont work if you drop some classes etc which in my case was not needed. Regards Rajeev J Sebastian
