Re: [xtm-wg] Graph again ... (and BTW, PSI again)

[Murray Altheim <altheim@eng.sun.com>]

Bernard Vatant wrote:
> 
> Murray
> 
> I just wanted to translate the feeling of the mathematicians, which are not
> completely mine, because I'm an uncomplete mathematician ... I am on the
> challenging task of bridging communities, remind you.
> My bridge to them was indeed that XTM syntax is what one can find the most
> simple as far as syntax is concerned, and since I could understand it and
> find it quite simple - as well as Steve and Michel's model for that matter -
> *they* could understand it.

And while I sometimes have difficulty with Steve's admittedly difficult
way of explaining himself, I find that when his thoughts are edited by 
someone like Sam I can understand it and it almost always makes sense.
Over the past few days I've been reading over the topicmaps.net PM and
while I'm also no mathematician, it makes sense to me. As I've said to
Steve and Michel, I think that since they've criticised Annex F as being
incorrect, they owe us an explanation of why, *in exchange* for us 
spending time reviewing their PM publicly. I'd much prefer they bring it
back into TopicMaps.Org, but I'm not wishing on the wind -- there's too
much work to be done and never enough time. But I must at least share 
my frustration at seeing a good piece of work get passed by, should the
current circumstances not permit us to work together effectively on such
important technical matters.

> I've followed and quite understood I think the reason for every bit of this
> syntax and agree with you on the feeling it's easy to use. And I'd gladly
> share the coffee before attacking any piece of RDF, but let me the entire
> pot, two or three cups will not do :)

I find it strange that after several years and even a private tutorial from
Guha, I still have difficulty with the syntax. At least the word is that
even the RDF people hate it, so I'm not alone. I can feel somewhat proud
that while XTM is admittedly verbose, word so far is that people like it.

> I agree a Topic Map is not "simply" a graph. But it's maybe very well
> represented as a "colored graph" or "labeled" graph. Which means the
> different kinds of nodes can be characterised by their "color" or "label" or
> "flavor" whatever you like to call them, like a-node and t-node and s-node
> ... and certainly other need to be introduced, so that we do not have graph
> inconsistencies like a node attached to a single arc, like in first analysis
> <roleSpec> appears to be, if you don't create a proper m-node. Things like
> that - more in the announced document.
> 
> Not in a corner on my mind the least intention to shed blood over the DTD.
> It should definitely be graph-representable as it is, like I think every
> possible DTD. Simply my hunch the graph will be more complicated that if it
> has been thought to begin with. That is : optimize and make consistent a
> graph structure out of the very core TM concepts, and then write the DTD and
> syntax out of it. Since we've made the road the other way round, forget
> about what-we-should-have-done.

Well, I learn always the hard way. My particular angle on a topic map engine
is one to build XTM documents. I'm less interested (at this point) in query,
etc. and more in maintaining more control over syntax, hence my keeping close
to the syntax in my Java classes. But as I've come to require some simple
merging functionality I'm finding both Annex F and Steve and Michel's current
PM document helpful in allowing me to think about both syntax constraints
and what should happen "in memory." Despite my respect for Eliot (which is
not I hope in question) I still have serious reservations about UML, simply
because I never grok it as well as the simple nodes-and-arcs diagrams I've
seen. Even the RDF spec uses such simple graphic language. As I mentioned
in an earlier message, I've still got the green marker diagram Eliot made
last year in Paris, some nodes and arcs centered around a binding point. I
hope we don't spend too much time making this more complicated than we 
need to.

> Concerning the "magic" algorithms. Mathematics are open source. Every bit of
> it. Is not that great ? Three centuries of work to get the Fermat's theorem
> demonstration, and not even a single license to use it. Well but ... I
> suppose implementations of those algorithms are not indeed open source, but
> I really don't know much about that. What I understand of what is
> implemented in Mondeca - which is very little, because I am not a developer
> at all - is that it was a huge and still ongoing work putting these
> algorithms into a proper effective "graph manager".

Well, I doubt I'd be able to use this since it's an academic project that
requires licensing for commercial use (although it never hurts to ask), 
but Dr. Dobbs this month discusses a series of Java packages called 
'The Data Structures Library in Java (JDSL)", which includes such
mathematical constructs as IncidenceListGraph, IntegerDijkstraPathfinder,
DirectedFindCycleDFS, etc. JDSL was developed by a group of mathematicians
and computer scientists, so it's probably pretty useful. I was looking 
over their javadocs to see what their graph implementations look like,
API-wise. Pretty cool. Too bad about it not being already open source.

> Anyway. I feel now that something has to get out of all that, but I'll be
> off that forum now till Saturday if I want to have my paper on PSI ready for
> Montreal submission. And BTW, the question I asked four days ago about PSI
> for PSI seems to have been swept away by the flow of messages about "the
> graph". Repeat : How do I do when I publish a PSI to declare it's intended
> to be one ?
> (see propositions in my previous message about it)

Well, I'm behind in writing up that proposal on publishing TopicMaps.Org
technical reports. You could write up a proposal for a PSI for PSIs and
submit it to the group for further discussion.

Murray

...........................................................................
Murray Altheim                            <mailto:altheim&#x40;eng.sun.com>
XML Technology Center
Sun Microsystems, Inc., MS MPK17-102, 1601 Willow Rd., Menlo Park, CA 94025

      In the evening
      The rice leaves in the garden
      Rustle in the autumn wind
      That blows through my reed hut.  -- Minamoto no Tsunenobu

------------------------ Yahoo! Groups Sponsor ---------------------~-~>
Do you have 128-bit SSL encryption server security?
Get VeriSign's FREE Guide, "Securing Your
Web Site for Business." Get it now!
http://us.click.yahoo.com/2cW4jC/c.WCAA/bT0EAA/2n6YlB/TM
---------------------------------------------------------------------_->

To Post a message, send it to:   xtm-wg@eGroups.com

To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com 

Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/