Re: [ba-ohs-talk] Towards a Graph API
Stephen Danic wrote:
> Graph repositories become highly intertwingular, so it's difficult to
pluck
> a "branch" and export it like you can with tree structures. (01)
Create a context or packaging node, and plug any other nodes
and/or relations
you want into it, then export? (02)
<shameless_plug>
http://www.concept67.fsnet.co.uk/gsix/
</shameless_plug> (03)
--
Peter (04)
----- Original Message -----
From: "Stephen Danic" <sdanic@memes.net>
To: <ba-ohs-talk@bootstrap.org>
Sent: Tuesday, March 05, 2002 6:19 AM
Subject: Re: [ba-ohs-talk] Towards a Graph API (05)
>
> > > Let's say we build an import/export standard. How do we make a
tree
> (XML)
> > > represent a graph?
> > > How do we prune a graph for exporting subsets, branches or
aggregates?
> >
> > GXL is a good starting point for an import/export standard. Of
course,
> > it's only a starting point; still have to develop the appropriate
GXL
> > schemas for different types of data.
> >
> > As for pruning a graph, what would be the Lucid Fried Eggs use cases
for
> > such a feature?
>
> Pruning would be used for sharing bits of data and synchronizing bits
of
> data. I currently keep all my notes in instances of Lucid Fried Eggs.
> Sometimes I would like to send a subset of nodes to someone else for
> inclusion in their database (sharing). I would also like to
synchronize a
> subset of nodes with Thoughtstream ( www.thoughtstream.org ) on a
handheld
> device.
>
> It's not essential. With email I can send people the text of one node
at a
> time, but it might be preferable to send them a little self-contained
> touchgraph ( www.touchgraph.com ) file.
>
> Imagine keeping lots of interrelational digraphs or ibis discussions
in a
> single repository, then exporting a single discussion and emailing it
to
> someone else.
>
> Graph repositories become highly intertwingular, so it's difficult to
pluck
> a "branch" and export it like you can with tree structures. We need
> mechanisms to select "aggregates of thoughts" ( www.ms.lt ) for
export. I
> suggest that algorithms for determining relatedness and selecting
aggregates
> should be included in the API. The algorithms themselves could be
simple or
> complex. A simple one is: all the nodes within x hops. A more complex
> algorithm might weight relatedness based on how many edges each
connected
> node has. Another algorithm might mine the textual content of the
nodes to
> determine relatedness.
>
> I've made another node at
> http://graphs.memes.net/index.php3?request=displaypage&NodeID=18 for
an API
> wishlist.
>
> ...Steve
> www.memes.net
>
>
> (06)