[crm-sig] Re: [crm-adhoc] Japanese aspects

martin at ics.forth.gr martin at ics.forth.gr
Wed Dec 5 09:59:38 EET 2001


Hi Tony,

I think we can separate the issue into two parts, the technical one and
the intellectual quality assurance.

>From the technical side, I can imagine the following:
An automated mapping mechanism is implemented, which exports the data in
into RDF. Formal validity of the mapping mechanism can be tested by 
validating a relevant sample (i.e. one which uses all entities and 
relationships of the CRM, which are used in the mapping algorithm).

This validation can be done e.g. by using a validating RDFS parser, which
would compare the created RDF instances with the CRM in RDFS form. 
Such parsers are available as freeware. 

I do not think, that creating RDF files or using XML Schema is a technology
widely available. Therefore I think creating instances of a "transport DTD"
 as the one we have used for the tests with the CIMI data, is 
more easy to perform. Once in this form, we have a trivial program we shall 
make available as freeware, that converts such data into the required RDFS
form. After that, anyone can run the validation test with an RDFS parser.

One question is, if we can regard the export capability into XML as widely
enough available. To my understanding, this can be performed by any 
database report writer, as the creating program needs not "understand" the
syntax of XML, but only blindly attach tags to contents.

Another question is, if we should recommend such export capability as an
important step to interoperability, and data sharing. If not exporting to
XML, what would be an alternative?

With respect to the quality aspect, an IT person could implement the 
export or mapping, together with the advice of the maintainers of the
museum information system. They would not understand the formalism of 
mapping, but they should be able to understand the logic of the produced
XML output, if in a human readable form, for instance via a standard XSL.
This XSL style sheet must be standard and validated by the CRM group, 
as XSL is so powerful, that is could "remedy" a wrong mapping.
Ther may also be a know-how question associated with the creation of 
exports/mappings

So one step is the agreement of the museum expert, that their data are in
transformed without change of meaning.

This poses another question, to which degree this interpretation conforms
with the intended use and meaning of the CRM as foreseen by the CRM Group.
I can imagine a validation of the latter by an expert. Experience must show,
what may be covered by good guide-lines and what may need human effort.

This poses the last question: how to institutionalize such procedures, and
how to scale up to the necessary availability of such a service.
This has aspects of training people, that want to offer validation service,
and it has an aspect of acceptance in the community, which form of service
is acceptable from an administrational point of view to the user 
organisations.

I think Nick can inform us about ISO standards for validation problems.

Another point is, that we see the CRM also as an intellectual guide. The
question is, to which degree a validation in such a role is useful and 
feasible, or in any other use of the CRM.

So far my thoughts.

best,

Martin

> Hi Martin,
> 
> I'm not quite sure what your asking for opinions about here...
> 
> Here are some things that sound like Good Ideas to me:
> 
> * Expressing the CRM as an XML schema;
> * Expressing the CRM as an RDF schema;
> * Exporting data from local systems as XML documents that comply with
> the  XML schema;
> * Exporting data from local systems as XML documents that comply with
> the  RDF schema;
> * Tools that can validate the XML documents against the XML or RDF
> schemas  (these already exist);
> * Stylesheets that allow the XML documents to be displayed in 
> human-legible form;
> 
> Is this what you were asking for opinions about?
> 
> Cheers,
> 
> T.
> ----------------------------------------------------------------------
> Tony Gill <> tony.gill at notes.rlg.org
> Research Libraries Group <> http://www.rlg.org/
> 1200 Villa Street, Mountain View, CA 94041 USA
> Voice: +1 (650) 691-2304 <> Fax: +1 (650) 964-1461
> 
> 
> 
> 
> 
> 
> 
> martin at ics.forth.gr
> Sent by: owner-crm-adhoc at ics.forth.gr
> 28/11/2001 08:57 PM
> 
>  
>         To:     crm-adhoc at ics.forth.gr
>         cc: 
>         Subject:        [crm-adhoc] Japanese aspects
> 
> 
> I had a presentation of the CRM at the Tokyo National Museum with
> participants from different organisations. There is strong interest. A
> key question was, to which degree one should restructure the existing
> information system to conform with the standard. I argued for the
> capability to export to a CRM compatible form. There was a strong
> interest in a validation service, which makes the issue of formally
> defining compatibility more urgent. I imagine a mapping process, which
> can be executed by an IT company, either as export with built-in tools
> of a  system
> or with an external tool (like data junction, etc.) into XML.
> We have a piece of code, that transforms XML with the simple CRM DTD
> into CRM RDF. This can formally be validated. The museum expert would
> validate the intellectual correctness of the mapping by reading the XML
> form with appropriate style sheet.
> 
> Opinions?
> 
> Cheers,
> 
> Martin





More information about the Crm-sig mailing list