[crm-sig] Software tools and the CRM
db at dbalzer.net
Mon Jul 8 17:29:43 EEST 2002
during the Copenhagen meeting Martin told me that there were
several people who doubt that the CRM expressed in RDF could
be processed by popular ontology editors such as Stanford
Protege-2000. Since this is my favourite tool for hacking
ontologies *including* the CRM, I think it might be useful if
I give a short overview of software tools I know of that can
be useful for modelling and application development using the
RDF version of the CRM.
All software packages are open source and freeware. Unless
otherwise noted, these are Java applications that require a
recent version of the JVM or JDK installed on your machine.
Suitable Java packages are available free of charge from
Sun or IBM.
Protégé-2000 from Stanford Medical Informatics (SMI)
This package has been around for several years, is actively
maintained, and has a fairly large user base. Many people have
contributed plug-ins that supply interfaces to other storage
formats and KR software packages, visualization, and data
Useful features that I have tried so far include a nice HTML
documentation generator, the visualization package, a
database storage backend using JDBC, and the facility to
combine and merge several (partial) ontologies. XML and Schema
support seems to be broken in the versions I have tried.
As a modelling tool, Protege allows you to extend the CRM by
creating sub-classes and sub-relations as well as instances.
Instances are kept neatly separated from class and property
definitions, e.g. by writing classes and properties to an
RDFS file while aggregating all instances in an accompanying
Protege loads all recent RDF versions of the CRM without
OilED from the University of Manchester
This is the current standard tool for editing DAML+OIL, an
RDF-based KR language. It claims to be the "NotePad of ontology
OilEd loads RDF versions of the CRM (make sure to import them
as RDF, not RDFS). To my taste, OilED is a bit more cumbersome
to handle than Protege when used as a CRM modelling tool. Models
can be verified with an external reasoning engine (FaCT) that
comes with the package.
Although I haven't played around with OilED to any significant
extent, I assume that this can be useful as a shuttle in and out
of DAML, a KR language in which an increasing number of
terminologies, thesauri and ontologies are being published.
OilED produces documentation in HTML similar to that obtained
from Protege. Export options include simple RDFS, a SHIQ
representation in XML from the Description Logics Implementation
Group (DIG), and a simple vector graphics format. All export
conversions seem to work well for the CRM.
IsaViz from the W3C
This is announced as a "visual authoring tool for RDF". It will
read the CRM in RDF and produce an awesome display of the entire
model. When it comes to authoring, however, anything more complex
than two or three RDF triples will be practically impossible to
handle via the current user interface.
Apart from the work done on RDF databases and query languages at
ICS-FORTH (on which Martin may be able to comment), there are
two packages that I find worth mentioning:
Redland RDF Application Framework from ILRT Bristol
This seems to be a solid library written in C with APIs for
various languages, including Java. A query language is not
included. However, there are several applications built on
top of Redland, some of which provide database and query
Jena RDF API from HP Labs, Bristol
For those doing application development in Java, this one seems
to be first choice. It comes with a very good tutorial. Working
my way through the examples went smoothly and, of course, it
digests the CRM in RDF without a hiccup.
Finally, for those looking for an RDF parser for Visual Basic,
there is a tiny package named Thea:
This relies on Microsoft's XML3 DLL which is notorious for having
difficulties with RDF syntax. CRM: no luck.
More information about the Crm-sig