Recovering Data from View
Download the following figure as OmniGraffle, PDF, or PNG.
GRDDL'ing the target graphical format file would produce some of the RDF that went into the visualization.
http://www.w3.org/TR/grddl-primer/#spreadsheets-section shows to use the following markup so that GRDDL agents can trigger and extract the RDF:
<BLAH
xmlns:grddl="http://www.w3.org/2003/g/data-view#"
grddl:transformation="https://raw.github.com/timrdf/vsr/master/src/xsl/grddl/graffle.xsl"
Applying grddl/graffle.xsl (page) produces a Turtle description of the graphical elements in an OmniGraffle file. In the sample below we see that a Rectangle with the label
http://danbri.org/foaf.rdf#danbri is depicting http://danbri.org/foaf.rdf#danbri (the flesh and blood walking around).
<1049>
a <ShapedGraphic>;
a <Rectangle>;
rdfs:label "http://danbri.org/foaf.rdf#danbri";
<http://open.vocab.org/terms/depicts> <http://danbri.org/foaf.rdf#danbri>;
<http://open.vocab.org/terms/originatingVisualArtifact> <http://logd.tw.rpi.edu/source/lebot-rpi-edu/dataset/visualizations/version/2013-Jan-24_13-47_1359053256264_ms>;
<http://open.vocab.org/terms/originatingVisualForm> <http://logd.tw.rpi.edu/source/lebot-rpi-edu/dataset/visualizations/version/2013-Jan-24_13-47_1359053256264_ms#visual_form_1048>;
<http://open.vocab.org/terms/vid> "1048";
<http://open.vocab.org/terms/vid_given> <http://danbri.org/foaf.rdf#danbri>;
<http://open.vocab.org/terms/vid_local_name> "#visual_form_1048";
<http://open.vocab.org/terms/visualContext> <http://logd.tw.rpi.edu/source/lebot-rpi-edu/dataset/visualizations/version/2013-Jan-24_13-47_1359053256264_ms>;
.
If each visualization generator captured the provenance of the visual that it made (we have this!), then we have a catalog of everything that is out there. With the catalog, we can grab each visual artifact, GRDDL them, and see that it does (or does not) depict the topic that we are looking for. We can now answer things like "Give me all visualizations that show Dan Brickley" with little more than a linked data crawler. The important thing here is that we do not control the entire system. Anybody can create visualizations about anything, and if they use the patterns that we describe, then anyone else can get this "topic indexing" for free without coordinating with everybody.
grddl.sh will look for grddl annotations and apply them to produce RDF from any file format.
See also: