A parallel, distributed algorithm for relational frequent pattern discovery from very large data sets