[crm-sig] Re:TOPOLOGY

martin martin at ics.forth.gr
Wed Oct 10 13:01:53 EEST 2001

Dear All,

I have invested some time to study the OpenGIS specifications, in
order to see, if the CIDOC CRM can learn something from them, but also 
to estimate what it would mean to feed data from CRM form to GIS

I started this exercise as a mapping, but finally only a very small
part can be mapped formally, not enough to make sense by itself. 
The reason is, that OpenGIS is mainly on another level. On one side,
it invests a lot in encoding and transformation semantics, using
mathematical geometric abstractions, on the other side it foresees
encoding of semantic information in the sense of the CRM, but without
declaring any such information, besides an "underspecified" concept
of "Feature". 

The way OpenGIS defines and deals with semantic information seems to be
completely compatible with the CRM, as well as the way the CRM deals with
measures seems to be completely compatible with OpenGIS, except for
accuracy handling.

Spatiotemporal topological operators as primary information like 
"overlaps with", "borders with", "before" "after" etc. we have to 
find from other sources, if we regard them as necessary.

They are quite well within the scope of OpenGIS, so OpenGIS could make
use of them, as well as of more elaborate "feature types".

The actuall overlap is in the relation between Place and Place Appellation,
where OpenGIS is compatible with the CRM, of course by far more detailed.
Similarly the notion of "measure" is compatible with CRM "Dimension".

The notion of time is more poorly defined in OpenGIS, and not completely
consistent throughout their documentation.

Accuracy handling in OpenGIS is solely based on statistical error models,
and solely based on interval arithmetic in the CRM. Both approaches make
sense and there are sound models how to relate both. As the CRM is not
made for computation-intensive reasoning, interval arithmetic is the better
solution.Besides that, historical data produce indeterminacy intervals rather than
statistical errors. To my opinion, OpenGIS should be dealing with both.

There could be a working group elaborating specifications about how 
cultural data could be interfaced with OpenGIS.

See my report:

 Dr. Martin Doerr              |  Vox:+30(81)391625          |
 Senior Researcher             |  Fax:+30(81)391609          |
 Project Leader SIS            |  Email: martin at ics.forth.gr |
               Centre for Cultural Informatics               |
               Information Systems Laboratory                |
                Institute of Computer Science                |
   Foundation for Research and Technology - Hellas (FORTH)   |
 Vassilika Vouton,P.O.Box1385,GR71110 Heraklion,Crete,Greece |
         Web-site: http://www.ics.forth.gr/proj/isst         |

More information about the Crm-sig mailing list