There are already projects that are open to all kinds of data (dbpedia comes to mind), and I don't think it is a good idea to replicate what they do.