Re: [Fenfire-dev] RDF

[Tuomas Lukka]

On Thu, Oct 30, 2003 at 03:23:19AM +0200, Matti Katila wrote:
> 
> Morning..
> 
> One question which has once and once again raised in my small head is RDF:
> local(ines) and invidual(ines). Currently we use cvs to store literary 
> rdf which seems to be temporary solution.

These are excellent discussion points. (BTW: If you work on one of these
enough, then that would easily go to a master's thesis or PhD, these
are pretty interesting questions...)

> - How user can have his/her own rdf graph *with* shared one?
>     - How to operate between personal rdf graph and shared/other one?

There are several possibilities here that haven't been thoroughly 
explored yet.

The "model_versions--tjl" PEG is related to this.

The possibilities include

1) explicitly adding to the RDF graph reified virtual triples for the source
   of each triple.

   i.e. we have the triple  a--b--c

   We would also have

   foo -- subject -- a
   foo -- predicate -- b
   foo -- object -- c
   foo -- sourcegraph -- osfjlosijflsijfe

2) merge the graphs and figure out when saving where to save what, by 
   simple regexp-like rules

3) Make our views and actions
   operate on several graphs at the same time explicitly

> - How to keep security between these two(or more)?

By not allowing scripting ;)

> - How to publish rdf graph or part of it?
>     - First idea is just to put storm pointer here but..?
>     - How to say - publish that canvas and all inside it?

Maybe a new graph with just those nodes, with *metadata* that says
"this graph is a subgraph of that graph". This could be made to
work with versioning, too!

> - What's our plan to displace web?
>     - If yes, how we are going to remedy the graphs?

You mean: "*Is* our plan to displace web"? Yes, certainly ;)

"remedy"?

>         - Is the UI of FenPDF the direction that it should look?

That is pretty hard to say... certainly some of the elements
but there are lots of different pressures over there...


        Tuomas