David Nickerson wrote:
>
> I'd suggest that since you are deciding how keywords are extracted,
> edited, and put back into the models that you are the best person to
> answer that. Otherwise there should have been some discussion on this
> mailing list as to how to handle this issue.
>
Okay, I've re-read the metadata specification, specifically the RDF
schema draft section at the bottom of the document, and the RDF schema
for Dublin Core (which was related).
<bqs:reference rdf:parseType="Resource">
<dc:subject rdf:parseType="Resource">
<bqs:subject_type>keyword</bqs:subject_type>
<rdf:value>
<rdf:Bag>
<rdf:li>ventricular myocyte</rdf:li>
<rdf:li>electrophysiological</rdf:li>
</rdf:Bag>
</rdf:value>
</dc:subject>
</bqs:reference>
This method of defining a keyword uses the straight dublin core subject
while using the bqs:subject_type property to label this instance of
dc:subject property as 'keyword', it is consistent with the standards
defined, but not specifically to the definition of the CellML metadata.
Perhaps those keywords were coded in before keywords were defined, and
nobody raised any issue about the need to convert the old format to the
new format until today.
If we use the metadata specification it may look like this:
|<bqs:reference rdf:parseType="Resource">|
| ||<bqs:keyword||>|||
| ||<rdf:Seq||>|||
|| <rdf:li>ventricular myocyte</rdf:li>
<rdf:li>electrophysiological</rdf:li>
||| ||</rdf:Seq>|||
| ||</bqs:keyword>|
|||</bqs:reference>|
||
Since the specification is already written to accomodate the keywords I
think it is better to keep to it, even though the common usage (as
defined by the number of files with this RDF to describe keywords used
by the old and current repositories) does not follow it. I must admit I
let this issue slipped through my head as I didn't think too much about
it. Thank you for correcting me.
However, my inspection of the metadata specification revealed some
issues, such as this:
| <!--Subject Type-->
<rdf:Property rdf:about="&bqsns;subject_type">
<rdfs:label>subject type</rdfs:label>
<rdfs:comment>
Defines the topic of a resource.
</rdfs:comment>
<rdfs:subPropertyOf rdf:resource="&dcns;subject" />
<rdfs:isDefinedBy rdf:resource="&bqsns;" />
</rdf:Property>
<!--Subject Type:Keyword-->
<rdf:Property rdf:about="&bqsns;keyword">
<rdfs:label>keyword</rdfs:label>
<rdfs:comment>
Defines the topic of a resource using keywords.
</rdfs:comment>
<rdfs:type rdf:resource="&dcns;subject_type" />
<rdfs:isDefinedBy rdf:resource="&bqsns;" />
</rdf:Property>
|
The specification have||| subject_type prefixed with both the dublin
core namespace and the bqs|| namespace. Dublin Core does not define
subject_type, so that needs to be corrected, perhaps the author meant to
use &bqsns;subject_type where s/he wrote &dcns;subject_type. It also
had double escaped ampersand (||&bqsns;subject_type should probably
be just ||&bqsns;subject_type) and other mistakes at other places. I
also think keyword be limited constrained to certain domains (via rdf
schema), such as limiting it to within the bqs:reference class so users
knows specifically where to place it, and machines know where to read it.||
I supposed this topic about formalizing the RDF schema and issues with
the metadata specification should probably be discussed in a different
discussion thread.|||
||
||Tommy.
_______________________________________________
cellml-discussion mailing list
cellml-discussion@cellml.org
http://www.cellml.org/mailman/listinfo/cellml-discussion
