Namespaces are good. But they generate a lot of URI's there just a little bit on the end is diffrent. I am tempted to try to split the URI in a prefix string and a name string (like in the table defenition above). That would give many nice effects: 1. The size of the database would be smaller. 2. Maby the indexes would be smaller/faster. 3. Just a prefix and no name would be taken to be a AboutEachPrefix (AEP) resource. 4. Much experimentation and visualization could be done with displaying only the name part. That is: I could create a tree for human reading, without having to wait for the support for RDF Schemas. But if I do that, I would have to append the prefix with the name in memory. Then, I can't do any search after a particular resource in the database. With the namespace 'www/' and name 'a/bc', I would have the resource 'www/a/bc'. But how can I find that this resource is covered by the AEP 'www/a/'?