2012-02-21

Protagoras and foaf:Person

πάντων χρημάτων μέτρον έστιν άνθρωπος,
των μεν όντων ως έστιν,
των δε ουκ όντων ως ουκ εστίν

About 2,500 years later, does the famous Protagoras aphorism still apply to things (χρήματα) in the Linked Data universe? The usual translation "Man is the measure of all things" could be misleading when put in parallel with the RDF motto "Everything is a Thing". The ancient Greek χρήμα has actually a more restrictive and pragmatic meaning, namely those things that one actually either holds and uses or badly needs, hence goods, possessions, money. And with this meaning in mind, the Protagoras aphorism is much less mysterious. Man(kind) is or should be the measure of what is needed and useful to mankind, either what one holds (things which are) or what one lacks (things which are not).
In the Linked Data universe, if everything is a thing, not everything is used or even useful.
But surely enough, many used and useful things are defined using reference to some human scale, as defined for example by FOAF. A broad look at the use of  FOAF either in LOD datasets or in linked vocabularies is already impressive. But looking more closely at the use and re-use of foaf:Person class in particular gets really awesome.
To figure it, go to the LOV Aggregator Endpoint and run the proposed default SPARQL query, or a more generic one such as the following.

DESCRIBE ?x 
WHERE{
{?x   ?p   foaf:Person}
        UNION
{foaf:Person   ?p   ?x}}
The queries are performed against a triple store aggregating the content of  more than 200 vocabularies (and counting) gathered in the Linked Open Vocabularies catalogue. The above general query will yield a raw aggregated RDF description of all "things" in this space directly related one way or another to foaf:Person.

  • Individual "vocab'lers" (?p = rdf:type) - creators or contributors to the vocabularies 
  • Subclasses and superclasses (?p = rdfs:subClassOf)
  • Disjoint ( ?p =  owl:disjointWith) and equivalent classes ( ?p =  owl:equivalentClass)
  • Properties attached one way (?p = rdfs:domain) or another (?p = rdfs:range)
All together those make quite a messy and certainly inconsistent graph, but nevertheless full of useful goodies, the  χρήματα of Linked Data measured by human scale. Do(es) your own data hold, use or lack them?