Sorry again for working my way around the limitations of this list
server. It is 2 am here, and I have spent a long time drawing these
graphs and trying to work out what the size limit of attachments are. I
won't do this again I promise :-/
On 11 Jun 2004, at 16:44, Karl Dubost wrote:
2 created Entry:
- published (issued)
Our BlogReader fetches the Atom feed and finds that the tag:e1 entry
has a publication date. It could at this point do one of two things.
a. It could create a new EntryState with exactly the same info as the
EntryState represented above, plus a issued node
b. It could just add the issued triple to the EntryState above
In both cases we will have a subgraph for the state like this.
The question will be: will the blank node x above be the same as the
blank node s in our initial diagram.
Things of course would be simple if s were not a blank node, but a
resource. Then one could immediately tell what to attach the new fact
to. But since the Atom syntax does not know of EntryStates, it is
impossible for it to pass that information along to us.
My guess is that for us both interpretations would lead to a truthful
model.
If it were possible to express a constraint of the form
for all Entries e, EntryStates es1, EntryStates es2, dates d
if e state es1 and e state es2 and es1 created d and es2 created d
then es1 == es2
and we agreed that this was true (because say it is impossible for any
human being to create two entries simultaneously, and only human beings
can create blogs)
then we could smush (technical term) node x and s in our graph.
