martin at ics.forth.gr
Wed May 15 14:23:53 EEST 2002
Still your example is not completely clear to me.
One aspect is, that a database record itself can be regarded as
a document. In that case, a "is documented in" link may connect
a node representing the database record with the nodes representing
the real world items referred to in it.
This is the interpretation we have given e.g. to the GENREG database,
where some records contain recorder information.
There is currently no link in the CRM, for which "is documented in"
represents directly a shortcut. A Conceptual Creation Event may
create a Document, which "takes into account" some othe information,
but there is no property of type
Currently, we can only state: Conceptual Creation created Document
Document documents Entity.
This allows to point to nodes, but not to links.
This restricts the granularity of reference to the sources. A reference
to a source for a link can only be simulated by a reference to the involved
But I'd like to stress, that the intention of the CRM is currently not
to support automatic conflict detection or resolution. The idea is, that
all assumed facts are registered (is produced by Production of axe 1,
is produced by Production of Set 1 of Axes), and summarize references to
the sources. Thereby we guarantee, that all relevant knowledge is gathered
at a unique point, but not, that sense is made out of it autmatically.
The latter functionality can easily be built on top of the CRM, but is,
to my understanding, a problem orthogonal to the semantic contents the
model describes, and as such out of scope.
The RDS reification construct is one, which points from a node (the source)
to a node-link-node tuple. As this is part of the Standard (RDF), it can
be used without need to become part of the CRM.
I would however discourage registering sources on such an "atomic" level.
The overhead is huge, and the benefit is a precision, which will hardly ever
be used by tools. Don't forget, that multiple opinions or alternatives
are often related to a very complex reasoning and estimates. To acquire
"yet another" opinion about who's right needs reading all arguments manually.
So, if I have a means to attach the sources to the next node, I think the
job of for the CRM is more than done. Source for the CRM is only a document.
The other argument was, that the optimal implementation of source reference
depends heavily on the number of sources involved, the frequency and
atomicity of the alternatives. Providing a "standard" solution is counter
productive. It would intervene with implementation issues, and does not
foster interoperability. Implementation suggestions for different optimal
implementations may quite well be given by the Group.
In general, one should find more about the problem under data warehouse
technology and "incremental data warehouse update". It is a problem
characteristic for data warehouses. Another source ar CSCW tools,
Computer Supported Cooperative Work, but they solve the dialogue management,
which is not the goal of the CRM.
Please let me know, if this answers your question.
If not, please make a short CRM instance demonstrating the point.
You could prepare something to be discussed in the next meeting.
Dr. Martin Doerr | Vox:+30(810)391625 |
Principle Researcher | Fax:+30(810)391609 |
Project Leader SIS | Email: martin at ics.forth.gr |
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/isl |
More information about the Crm-sig