self-join-2.pig

edges     = LOAD 'graph.tsv' AS (v1:chararray, v2:chararray);

--
-- Augment the edges with the sizes of their outgoing adjacency lists.
--
grouped_edges = GROUP edges BY v1;
aug_edges     = FOREACH grouped_edges GENERATE FLATTEN(edges) AS (v1, v2), COUNT(edges) AS v1_out;
aug_dups      = FOREACH aug_edges GENERATE v1, v2, v1_out;
--
-- Compute the sizes of the intersections of outgoing adjacency lists
--
edges_joined  = JOIN aug_edges BY v2, aug_dups BY v2;
intersection  = FOREACH edges_joined {
                  --
                  -- results in:
                  -- (X, Y, |X| + |Y|)
                  -- 
                  added_size = aug_edges::v1_out + aug_dups::v1_out;
                  GENERATE
                    aug_edges::v1 AS v1,
                    aug_dups::v1  AS v2,
                    added_size    AS added_size
                  ;
                };
DUMP intersection;