[Nepomuk] Brainstorming: Metadata Sharing

[Sebastian Trüg]

not crazy at all. This was the plan in the original Nepomuk research
project. Only they never managed to make it work properly.

On 06/28/2010 10:52 PM, Artem Serebriyskiy wrote:
> Some crazy idea( really crazy ):
> What about using some kind of p2p network for sharing metadata ? It
> might be usefull for sharing public metadata - for example information
> about music, about books, about films.
> 
> On Mon, Jun 28, 2010 at 11:12 PM, Vishesh Handa <handa.vish at gmail.com
> <mailto:handa.vish at gmail.com>> wrote:
> 
>     Hello Oszkar
> 
>     On Tue, Jun 29, 2010 at 12:14 AM, Oszkar Ambrus <aoszkar at gmail.com
>     <mailto:aoszkar at gmail.com>> wrote:
> 
>         Hi Vishesh,
> 
>         On 28 June 2010 18:50, Vishesh Handa <handa.vish at gmail.com
>         <mailto:handa.vish at gmail.com>> wrote:
>         > Metadata sharing implies that we should be able to see other
>         people's
>         > metadata and have it on are own system. Sebastian suggested
>         that we store
>         > user information ( who owns which statements ) as graph
>         metadata. That's
>         > totally feasible, and maybe we could also store the permission
>         settings as
>         > graph metadata.
> 
>         I'm thinking of permissions settings similar to how you share stuff
>         through Samba.
>         You can share your stuff with guests (i.e. everyone, without
>         authentication), with all known users (stored locally) or with a
>         subset of existing users.
>         The list of users could be stored locally as RDF and then, similarly
>         to what Sebastian suggested, statements can be annotated with the
>         graph meta-data to specify privileges.
> 
> 
>     Yes. I had something similar in mind.
>      
> 
> 
>         > I'm not too sure how we would choose whose metadata to store
>         our on system
>         > or why we would need to do that.
> 
>         As for whose meta-data to store, would you like to go with a
>         high-level protocol, such as Jabber? So no low-level stuff?
> 
> 
>     Forget about Jabber. I think it would be better to go with
>     Telepathy. Daniele [0] is working on "Telepathy Tubes and File
>     Transfer in KDE", and it would be a lot better to support multiple
>     protocols via Telepathy instead just supporting Jabber. Once his
>     project is complete we should be able to export a Dbus interface to
>     other contacts, so that greatly simplifies the problem of how to
>     connect/transfer stuff between 2 machines.
> 
>     Please look at the attached conversation.
>      
> 
>         Because, as I understood, the RDF repository is going to be exposed
>         through HTTP, so you could connect to anyone's repository and just
>         insert their address in the graph metadata of the statements you
>         store
>         locally.
> 
> 
>     No, I don't think we should expose it via HTTP. ( Might be
>     problematic ) There is a project in the playground called nsqd
>     (Nepomuk Social Query Daemon ), which does something similar. I'll
>     take a look at it.
> 
>     - Vishesh Handa
> 
>     [0] http://blogs.fsfe.org/drdanz/
> 
>     _______________________________________________
>     Nepomuk mailing list
>     Nepomuk at kde.org <mailto:Nepomuk at kde.org>
>     https://mail.kde.org/mailman/listinfo/nepomuk
> 
> 
> 
> 
> -- 
> Sincerely yours,
> Artem
> 
> 
> 
> _______________________________________________
> Nepomuk mailing list
> Nepomuk at kde.org
> https://mail.kde.org/mailman/listinfo/nepomuk