Application of Filters to Multiway Joins in MapReduce

Algorithm 5

Reduce function.

 ⊳𝑣𝑎𝑙𝑢𝑒𝑠: intermediate records (𝑟𝑒𝑐𝑜𝑟𝑑,𝑑𝑎𝑡𝑎𝑠𝑒𝑡𝐼𝑑) with the same reducer id

(1) procedure  REDUCE(𝑘𝑒𝑦,𝑣𝑎𝑙𝑢𝑒𝑠)

 ⊳𝑟𝑒𝑐𝑜𝑟𝑑𝐿𝑖𝑠𝑡: lists for buffering the intermediate records based on their dataset id

(2)  𝑟𝑒𝑐𝑜𝑟𝑑𝐿𝑖𝑠𝑡←∅

(3)  for  each 𝑣𝑎𝑙𝑢𝑒 in 𝑣𝑎𝑙𝑢𝑒𝑠  do

(4)   𝑑𝑠𝑖𝑑←𝑣𝑎𝑙𝑢𝑒.𝑡𝑎𝑔

(5)   𝑟𝑒𝑐𝑜𝑟𝑑𝐿𝑖𝑠𝑡[𝑑𝑠𝑖𝑑]←𝑟𝑒𝑐𝑜𝑟𝑑𝐿𝑖𝑠𝑡[𝑑𝑠𝑖𝑑] ∪ 𝑣𝑎𝑙𝑢𝑒.𝑟𝑒𝑐𝑜𝑟𝑑

(6)  end for

 ⊳𝐽𝑜𝑖𝑛(𝑟𝑒𝑐𝑜𝑟𝑑𝐿𝑖𝑠𝑡): returns the join results between the records in each record list

(7)   𝐸𝑚𝑖𝑡(𝐽𝑜𝑖𝑛(𝑟𝑒𝑐𝑜𝑟𝑑𝐿𝑖𝑠𝑡))

(8) end procedure