[Om3] AfterOpenMath: a new proposal for extensible metadata in CDs

[Christoph LANGE]

Dear all,

  thinking about those semantic web CDs I promised to contribute (i.e. the
ones we've first written in OMDoc, for that "ontology" paper at MKM), an idea
came to my mind how we could add arbitrary metadata to OpenMath CDs.

As a part of the new way of handling metadata in OMDoc, we also allow for
attributing mathematical (sub-)expressions with metadata, using OMATTR.  This
becomes possible, as we treat all metadata vocabularies (= ontologies) as CDs.
For example:

<OMOBJ>
  <OMATTR>
    <OMATP>
      <OMS cd="dc" name="description"/>
      <OMSTR>now comes a really interesting application of 'plus'</OMSTR>
    </OMATP>
    <OMA>
      <OMS cd="arith1" name="plus"/>
      <OMATTR>
        <OMATP>
          <OMS cd="dc" name="contributor"/>
	  <!-- not sure if I may abuse OMR to point to 
	       arbitrary URIs, so we might think about
	       an alternative here -->
  	  <OMR href="http://kwarc.info/clange"/>
	  <OMS cd="rdfs" name="comment"/>
	  <OMSTR>I thought that '1' would make the best example here</OMSTR>
        </OMATP>
        <OMI>1</OMI>
      </OMSTR>
      <OMI>1</OMI>
    </OMA>
  </OMATTR>
</OMOBJ>

This would ultimately allow us to avoid discussions as in
https://trac.mathweb.org/OM3/ticket/12, as we could easily make a CD refer to
arbitrary metadata by attribution.  (Well, yes, I know, the respective
applications would still have to support those metadata CDs, but on the other
hand, in the 'compliance' section we have a well-defined mechanism for
negotiating support among applications.)

In a CD, it could look like this:

<CDDefinition>
  <Name>sym</Name>
  <Description>foo bar</Description>
  <!-- standard metadata end here -->
  <FMP>
    <OMOBJ>
      <OMATTR>
        <OMATP>
	  <OMS cd="dc" name="creator"/>
	  <!-- if unambiguous we can use username strings or 
	       real names here, but URIs would be better, see above -->
	  <OMSTR>jhd</OMSTR>
	  <!-- further metadata go here -->
	</OMATP>
	<OMS cd="mycd" name="sym"/>
      </OMATTR>
    </OMOBJ>
  </FMP>
</CDDefinition>

I.e. one would attribute the single symbol with the desired metadata.  This
would relieve us from some pressure concerning the extension of the OCD XML
schema itself.  We might see what metadata are most frequently used in the
"attributed" syntax, and then eventually introduce proper XML elements as
syntactic sugar for the most frequently used ones.

And, in the interest of semantic web compatibility, we should also specify the
existing metadata in terms of the attributed syntax, e.g. mapping
<Description/> to dc#description.

Note that FMP might not be the right thing to use here, as it would blur the
difference between data (= proper FMPs) and metadata (= this style of FMP).
But as we are at introducing different types of FMPs anyway (DefMP), why not
introduce sth. like <FMP type="metadata"> ?

What do you think about this? -- Happy to hear your feedback.

Cheers,

Christoph

-- 
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
Url : http://openmath.org/pipermail/om3/attachments/20090716/07c295d0/attachment.pgp