Re: [xdi] notes from meeting in Nice, Jan 29 2009

[Nick Nicholas <opoudjis@optushome.com.au>]

Thank you, useful to get that level of explicitness. I just discovered  
today reviewing the RDF model that = is itself an XRI for the =  
registry, so that =drummond => =/$has/drummond : your model addresses  
this nicely.

You're not making any commitment about what the relationship is  
between +x+y and +y. I had been suggesting +x+y/$is$a/+y, but after  
revisiting the RDF model (and statements such as $get$a$xsd$boolean),  
I'm not as confident about that, so you needn't change anything.

The model does not relate $has to $has$a. I think there is clearly  
such a relation --- the object of $has is the unique, possibly  
aggregate value, the object of $has$a includes those values in classes  
--- and it would be nice to have a diagram showing the relation  
between =abraham/$has$a/+son and =abraham/$has/=isaac. The  
relationship of $is and $is$a, on the other hand, was always somewhat  
murky, and it would not be useful to attempt to illustrate it.

In the new dispensation, the inverting word is $is, not $a. :-)

On 30/01/2009, at 13:15, Markus Sabadello wrote:

> Hi Giovanni (and others),
>
> I'm trying to summarize in a few words what we talked about. Please  
> correct me if I made a mistake or forgot anything.
>
> - We had a look at the xdi-rdf-global-graph-proposal-v1 which  
> Giovanni posted about a week ago:
> http://www.oasis-open.org/committees/download.php/30802/xdi-rdf-global-graph-proposal-v1.pdf
>
> - The basic graph model described in that document is essentially  
> the same as the one we are all familiar with (i.e. subject/predicate/ 
> object triples), but the document proposes a new methodology for  
> drawing the graph. The main innovation is to draw EVERYTHING that  
> has an XRI as a node (including predicates!). Arcs are drawn between  
> the nodes, but arcs themselves don't have XRIs. A typical statement  
> consisting of subject, predicate and object would be drawn as three  
> nodes and three arcs.
>
> - Arcs are colored to indicate which statement they belong to.
>
> - A node with a given XRI never appears twice in the graph, but it  
> can of course be part of more than one statement. For example if you  
> have the following two statements:
> =markus/+email/'markus.sabadello@gmail.com'
> +email/$is$a/+contactdata
>
> Then only a single node with the XRI +email would be drawn, but it  
> would be part of both statements.
>
> - Giovanni distinguishes between Syntactic Correctness and Semantic  
> Consistency
>
> - Syntactic Correctness simply means that the graph properly fits  
> into the RDF graph model (subject/predicate/object). This is a  
> similar concept to a "well-formed" XML document.
>
> - Semantic Consistency means that the statements in the graph are  
> consistent with accompanying dictionary statements. E.g. if you have  
> a statement =markus/+friend/=giovanni, then for the graph to be  
> "semantically consistent", there would also have to be the  
> dictionary statement =markus/$has$a/+friend. These dictionary  
> statements could be mixed into the same graph, or they could exist  
> in a separate document or XDI endpoint (such as a Community  
> Dictionary Service). This is a similar concept to a "valid" XML  
> document.
>
> - The XDI Validator application could be extended to support  
> validating this Semantic Consistency.
>
> - If you want you can work with XDI without caring about "semantic  
> consistency" / dictionary entries, just like you can work with XML  
> without caring about schemas / DTDs.
>
> - Giovanni's document also introduces the concept of "roots", which  
> are drawn as a slash in a box. Those roots and the arcs originating  
> from them do not actually appear as statements in the graph, but  
> they indicate who is authoritative for an XRI. For example, all +  
> XRIs (+friend, +email, etc) come from the same "root" (the Community  
> Dictionary Service). Similarly, all $ XRIs ($is, $has, etc) come  
> from the same "root" (the XDI specfications). =markus and =drummond  
> would come from different "roots", since different XDI endpoints are  
> authoritative for them.
>
> - We talked a bit about equivalence. Giovanni thinks that if there  
> is a $is statement between two XRIs, then those two XRIs are drawn  
> as only a single node in the graph. So the node would have two XRIs  
> that identify it. Markus mentioned that is not how his  
> implementation currently works.
>
> - A question that follows is how would XDI messaging be affected by  
> the above. For example, if you have the following XDI graph:
>
> =markus
>     $is
>         =markus.sabadello
>     +name
>         "Markus Sabadello
> =markus.sabadello
>     +nationality
>         "AT"
>
> And if you then send this XDI message:
>
> =giovanni
>     $get
>         /
>             =markus
>                 +nationality
>
> Would there be a reply ("AT"), or would there be no reply?
>
> - A similar example would be the effect of a $del message. If  
> equivalent XRIs are drawn as only a single node in the graph, and  
> you remove the $is statement via a $del message, would the node then  
> "split up" into two nodes?
>
> - Giovanni gave a demo of a project his team has been working on  
> ("SMS" - "Simple Mobile Services" - www.ist-sms.org). It allows  
> users to send small notes about people, locations, services, web  
> sites, etc from their mobile phone to a server. These notes (so- 
> called MEMs) are then shared with an online community. It's possible  
> to advertise and discover new services. Phone security features such  
> as encryption via the SIM card are used. Various authoring tools are  
> available for creating new Simple Mobile Services that communicate  
> via MEMs. A web portal makes it possible to collect and analyze the  
> data that was collected by the system.
>
> - Giovanni explained that many of his XDI inputs were motivated by  
> use cases in the SMS system, e.g. XDI Queries.
>
> - We both think it was a good idea to have a chatroom during the  
> last TC call. We should always use that from now on.
>
> Markus
>

(*)(*)(*)
http://www.opoudjis.net      DR NICK NICHOLAS; LINK AFFILIATES,  
MELBOURNE
skype:opoudjis              opoudjis@optushome.com.au
  "Some of the English might say that the Irish orthography is very  
Irish.
  Personally, I have a lot of respect for a people who can create  
something
  so grotesque." -- Andrew Rosta