Dear all
In Rethymno the program was tight and there was no time for me to give a
presentation of what we are doing for the museums and collection
departments at the Norwegian University Museums. We have two power point
presentations describing a project we are doing for the archaeological
museums especially in Bergen and Trondheim and a more general presentation
of the user applications respectively. The latter is a translation of a
demonstration-presentation of our software given a language technology
conference for Norwegian language. The first part is general and also
museum oriented. The second part is oriented towards corpus linguistics and
lexicography and may be not of general interest.
Archaeology: http://www.dok.hf.uio.no/nedlasting/foredrag/topark.ppt (2.5
Megabytes)
General http://www.dok.hf.uio.no/nedlasting/foredrag/com_appl.ppt (1.6
Megabytes)
The work described in the first presentation is in fact mappings from three
different document collection types into CRM-like model(s). The model is
inspired by the work of Lene Roll and her colleagues at Nationalmuseet in
Copenhagen back in 1993. These exercises show that it is possible to
extract quite a lot of information from old documents by doing mechanical
XML/SGML markup. However, if one wishes to populate a more complete CRM
model, one has to perform an extensive interpretation of the content of the
documents. This is extremely expensive for the quanta we are dealing with
and beyond our funding.
In a full CRM-model the tagged full text catalogues should be strings
connected to documents documenting the events. In our discussion in
Rethymno Martin argued that if a database is CRM compliant, no structured
information should be "left" in the free text fields. We have extracted as
much as possible of the information from the texts, but still we wish to
keep the text with the markup as an integrated part of the database.
As the second presentation illustrates, we have an object oriented layer or
model on top of the relational structure. The object oriented layer is
implemented in a separate relational table structure called The Meta
Database which are interpreted run time by the user application(s). We are
in the process of adjusting the model in the direction of CRM. In the Arena
project (http://ads.ahds.ac.uk/arena) we did not succeed in convincing our
partners that CRM should be the basis, thus we will make a run time mapping
from our model to the simpler data exchange interface.
I hope some of this may be of some interest.
Best wishes
Christian-Emil
