Social Relationships and Ontologies

29Jul07

Clay Shirky brings up an interesting point in his post about describing relationships with a controlled vocabulary. Clay argues that a formal and explicit ontology for human relations is unworkable from a philosophical aspect – that is relationships between people cannot be neatly boxed into predefined categories. I tend to agree…

However, defining such relationships is exactly what social software services does - such as LinkedIn that e.g. define the relationships ManagedDirectlyBy and WorkedInGroupWith. When that is the case and we want to express the relationships in RDF we will need an ontology which is exactly what RELATIONSHIP does.

It is safe to assume that services will require different ontologies as each service has different requirements. As Clay points out a single ontology defining any type of relationship is just not possible, but using RDF this is easy as one can use whatever ontology that may match the requirements.

For this to be useful and allow systems to interoperate we must link between ontologies. Doing so may be successful in areas such as biology, medicine and genomics where concepts are unambiguous, but how do we link between abstract relationships? Can WorkedInGroupWith be linked to ColleagueOf or the WorksWith property of RELATIONSHIP? Not likely – In fact the only safe link may be to the property KnowsOf in which case most meaning will be lost.

In my opinion there is no solution to this problem and it really illustrates the weakness of semantic web technology – It assumes a simpler world but the fact is that relationships are not simple. In fact relationships mean different things to different persons and cultures – that is there is no shared worldview. Take the property ColleagueOf. Due to the Power Distance of the Danish culture Danes may think of their manager as a colleague while in China this is unheard of.

Good luck making that work…

Advertisement

Filed under: Social Software, The Semantic Web   |  Leave a Comment