Reduce and aggregate: similarity ranking in multi-categorical bipartite graphs

   Abstract