Efficient and Scalable Graph Similarity Joins in MapReduce

Algorithm 2

Replacement of filter of Algorithm 1.

(1) Map   ⟨𝑟𝑖𝑑,𝑟⟩/⟨𝑠𝑖𝑑,𝑠⟩→𝑟𝑖𝑑,𝑠𝑏𝑓_𝑟/𝑠𝑖𝑑,𝑠𝑏𝑓_𝑠

(2)   create SBF for each graph and insert the generated 𝑞-gram signatures into it;

(3)   output 𝑟𝑖𝑑,𝑠𝑏𝑓_𝑟 or 𝑠𝑖𝑑,𝑠𝑏𝑓_𝑠;

(4) Shuffle conduct Cartesian product between pairs 𝑟𝑖𝑑,𝑠𝑏𝑓_𝑟 and pairs 𝑠𝑖𝑑,𝑠𝑏𝑓_𝑠;

(5) Reduce   𝑟𝑖𝑑,𝑠𝑏𝑓_𝑟,𝑙𝑖𝑠𝑡𝑠𝑖𝑑,𝑠𝑏𝑓_𝑠→⟨𝑟𝑖𝑑,𝑠𝑖𝑑⟩

(6)   for any ⟨𝑟𝑖𝑑,𝑠𝑖𝑑⟩, calculate the intersection of 𝑠𝑏𝑓_𝑟 and 𝑠𝑏𝑓_𝑠 and estimate

     the number of common signatures;

(7)    invoke the count filtering for each pair ⟨𝑟𝑖𝑑,𝑠𝑖𝑑⟩ and output it if it passes the condition;