This is code to compare SPARQL query results to each other. The results that
have a low similarity to the rest of the set will be discarded as outliers. The
goal is to clean up noisy result sets.

We use 16 queries with enumerations (types of curry, tupes of kangaroos, etc).
These have been graded by a human, so they are our gold standard. The queries
are run against factForge.