Context: 

 * http://downlode.org/rdf/file-properties/ (by Earle Martin)
 * http://chatlogs.planetrdf.com/swig/2006-04-25.html#T18-01-57

Current approach: 

  <fp:checksum>
    <fp:Checksum>
      <fp:generator>
        <wn:Algorithm rdf:about="http://en.wikipedia.org/wiki/MD5" />
      </fp:generator>
      <fp:checksumValue>e89760b5510974acbea8ff4c77a1a5c0</fp:checksumValue>
    </fp:Checksum>
  </fp:checksum>

Flat approach, suggested by theorbtwo on #swig: 

   <fp:md5>e89760b5510974acbea8ff4c77a1a5c0</fp:md5>

Datatypes approach: 

   <fp:checksum rdf:datatype="http://example.org/datatypes/md5"
                >e89760b5510974acbea8ff4c77a1a5c0</fp:checksum>

Interpretation properties approach: 

   <fp:checksum rdf:type="Resource">
      <fp:md5>e89760b5510974acbea8ff4c77a1a5c0</fp:md5>
   </fp:checksum>

Note how the flat, datatypes, and interpretation approaches all incur significantly less overhead than the current approach; but as to which of them is preferable, it's difficult to say.

Nuts, make that: 

   <fp:checksum rdf:parseType="Resource">
      <fp:md5>e89760b5510974acbea8ff4c77a1a5c0</fp:md5>
   </fp:checksum>