[topicmapmail] [ANN] Topic Map Explorer Version 1.0
rick coope
rick.coope@tmexplorer.com
Tue, 7 Sep 2004 14:16:34 +0100
Hi Jan,
Regarding basic concepts, perhaps core concepts is more appropriate, we see
Topics as providing the binding point for identities of subjects,
associations define the connections between topics, and occurrences, while
really only special associations, relate topics to information resources.
This is captured in XTM in the topic, association and occurrence elements,
in the TMDM in the related Information Items and in general in the
literature regarding topic maps.
While in model terms we see that things can be reduced to a lot fewer
structures, we think that message wise having Topics, Associations and
Occurrences as the building blocks is a useful vehicle for communication and
explanation.
I agree with you that merging is one of the basic concepts but it needs
those structures to have any meaning.
As for support for merging, there are three different kinds of merging that
occur with TME, two of which are exposed in the user interface and a third
that isn't yet. The two that are exposed are name based merging, if you
create a topic with the same name as one that already exists then you can
choose for them to become the same topic or to have different identities. If
you merge two topics all associations and properties are merged onto one new
topic.
The second merging capability is based on topic identity and is exposed in
the p2p search capabilities. P2P search is initially by name, it could also
be by PSI, but for now it is by name. However, once topics are returned
subsequent p2p searches, to find out more about a topic, are done by
identity and all properties relating to the same topic on different peers
are merged and displayed to the user.
In the model, the RDF identity of a resource is basically a system assigned
src locator. There are RDF predicates to indicate subject indicators and
resource references. These will be exposed in future releases at the user
interface level (topic property editor), p2p searching and XTM import.
Regards,
Rick
www.tmexplorer.com
-----Original Message-----
From: jan [mailto:jan] On Behalf Of Jan Algermissen
Sent: 07 September 2004 10:31
To: rick coope
Cc: 'Thomas B. Passin'; topicmapmail@infoloom.com
Subject: Re: [topicmapmail] [ANN] Topic Map Explorer Version 1.0
Hi Rick,
rick coope wrote:
> The Topic Mappy ness issue. As I mentioned in the rdf interest list
> posting, we are trying to embody the basic concepts of Topic Maps,
> (topics with identity, associations and occurrences).
For various reasons, related to my own work, I am curios what parts of
ISO13250 or XTM 1.0 lead you to believe that Topic Maps's 'basic concepts'
are (topics with identity, associations and occurrences) and what your
understanding is of how these should be implemented to create a 'Topic Map
application'.
I personally have allways understood (regardless of how you approach Topic
Maps!!) that the basic concept is 'merging', especially, that after an 'act
of data integration' (merging diverse data sources) all information about
something will be (virtually) accessable from a single point.
Does Topic Map Explorer support this?
Jan
There are of course limitations in
> representing things as a tree and as you and Murray have discussed
> this is a design decision. We felt at this stage that representing
> things as a tree was familiar to users but that the added level of
> adding the relationship type made useful groupings and added value and
aided navigation.
>
> We will support XTM but more importantly we will support TMQL and a
> topic map remote access protocol as these are provable and testable
things.
>
> Things such as PSIs are not yet exposed in the user interface but are
> supported in the model and are used in the peer-to-peer capabilites.
> Again scope is not something we were clear how to expose to the user
> and so didn't for now.
>
> Although you made a loop you didn't want to, you can correct that
> simply by selecting the topic and pressing delete. You could not set a
> topic type as probably you had not created any.
>
> Tom, we do appreciate your comments and there are things that need to
> be addressed however I feel you have been overly harsh given you
> haven't read the documatation that accompanies the product (we need to
> make it clearer where to find this) and thus you havent used it in the
> way it is intended and thus not gained the benefit it is designed to
> deliver. Even simple operations like deleting a relationship or
> assigning a topic type, have thus caused problems.
>
> Regards,
>
> Rick
>
> Tom Passin wrote:
>
> I have downloaded this application and tried it out for all of ten
minutes.
> I am disappointed to report the following -
>
> 1) This is not a production quality release, event though it is styled
> "verson 1.0". I got numerous exceptions (which were at least
> recoverable) and many anomalies, including a mislabled dialog component.
>
> 2) This is not a topic map application in any way, shape, or form. It
> does not export or import any recognized topic map format. Instead it
> exports rdf! But it does not import general rdf. Or rather, on the
> simple rdf files I tried, it claimed success, but the display showed
nothing.
>
> The designer appears not to understand the basics of topic maps.
> There is no way to indicate the subject of a topic. There are no scopes.
> There are only binary relations. The only display is a tree view.
> There are no PSIs. I was unable to delete a relation I had created.
> I could not find a way to change a relationship type once one had been
> created. You are supposed to be able to designate the "type" of a
> topic, but that didn't work. You can't relate property types to each
other.
>
> I was also able to (unintentionaly) create an loop between two topics
> I tried to relate, so that the tree view display kept unfolding the
> same two topics ad infinitum (if I had wanted to keep expanding the tree).
>
> Of course, the fact that there was no documentation probably didn't help.
> Still, it is *inexcusable* to call this a "topic map explorer"
> and not have a *single* characteristic that disinguishes topic maps
> and makes them rich and valuable.
>
> Setting aside the issue of whether this application has anything to do
> with topic maps per se, could it be useful in and of itself? Setting
> aside current bugs and questionable interface features - which are
> numerous - in my view, not with just a tree view, and not with the
> apparent limitations of the internal model. I say "apparent
> limitations", because it is possible that the internal model is richer
> than the app exposes (maybe there can be a hierarachy of association
> types, for examples), but one could not know by the current interface.
>
> What really bothers me is that this app could give topic maps a bad
> name, or at the least mislead people into thinking that topic maps are
> very limited and don't do anything that hasn't been seen in many apps
before.
>
> I almost feel like I should apologize to the developer of this product
> for my strong words, but I mean them all. Whoever you are, this
> product would have to go a long way before I would even consider
> trying another evaluation copy. I don't even feel like encouraging
> you to try to modify and extend it, because the current starting point is
so weak.
>
> I hope I am wrong about my last statement, because I would really like
> to see a modestly priced, easy to use topic map application.
>
> Regretfully yours
>
> Tom P
>
> --
> Thomas B. Passin
> Explorer's Guide to the Semantic Web (Manning Books)
> http://www.manning.com/catalog/view.php?book=passin
> _______________________________________________
> topicmapmail mailing list
> topicmapmail@infoloom.com
> http://www.infoloom.com/mailman/listinfo/topicmapmail
>
> _______________________________________________
> topicmapmail mailing list
> topicmapmail@infoloom.com
> http://www.infoloom.com/mailman/listinfo/topicmapmail
--
Jan Algermissen http://www.topicmapping.com
Consultant & Programmer http://www.gooseworks.org