As I prepare for the next meeting, I have been
thinking about my personal biases as they may affect any contributions I may
make to the XML Topic Maps effort. I propose to lay them out briefly here
so that others may ponder them and respond before or at the meeting. My
biases, as expressed here, should not be considered of the *strong* (read:
disruptive) nature; rather they represent my personal desire to make Topic Maps
perhaps something more than that for which they are intended.
I take a strong interest in pedegogy and the use of
various tools by which humans construct their internal representations of
knowledge. Topic Maps strike me, at least at this moment, as an indexing
scheme, one which provides views constructed by the user to suit his/her
particular needs. I do not presently regard TMs as a knowledge
representation scheme, rather an index or view into them.
Standing alone, without recourse to occurances, a
TM appears much like a Concept Map ( http://www.to.utwente.nl/user/ism/lanzing/cm_home.htm ).
But, as I presently understand them, TMs build a
kind of semantic network with pointers into documents of other types such as
text files. These other documents, to me, represent a record of human
dialog. Thus, I see a two-layer structure: index and dialog.
I tend to wish for a third layer, one interposed
between the TM layer and the dialog record. That layer would be, as I see
it, a much richer representation scheme, perhaps along the lines of Conceptual
Graphs (John F. Sowa's rendition of Peirce's Existential Graphs) ( http://www.cs.uah.edu/~delugach/CG/ )
or OML/CKML
or others as suggested at http://www.oasis-open.org/cover/OntologyExchange.html
My reason is driven by the notion that the recorded
dialog, itself, is often not usable for direct machine inference, and that TMs
may not retain the necessarily fine granularity in expression of relationships
and taxonomies necessary for inference. I believe that such a rich knowledge
representation layer can be constructed from the dialog record, and that TMs
might offer, in addition to the occurance links, specific links into nodes in
the middle layer.
I therefore wonder to what extent the XTM committee
shares any part of my thinking. I look forward to such a
discussion.
Jack Park
To Post a message, send it to: xtm-wg@eGroups.com
To Unsubscribe, send a blank message to: xtm-wg-unsubscribe@eGroups.com