wrong result for BipartiteSimRank

[antahiap]

testing the BipartiteSimRank for the paper example, showing the node 0-1 is less similar than 2-3 however it should be opposite.

source target weight
0 0 4 1000
1 1 4 1000
2 2 4 1
3 3 4 1

Percent: [##############################] 100% Complete!
Converged at iteration 2

          0             1       2       3

0 1.000000e+00 8.000000e-07 0.0008 0.0008
1 8.000000e-07 1.000000e+00 0.0008 0.0008
2 8.000000e-04 8.000000e-04 1.0000 0.8000
3 8.000000e-04 8.000000e-04 0.8000 1.0000

[mrhhyu]

Did you run the provided matrix-based source code?
this matrix-based implementation of SimRank is fully tested and compared with the original loop-based implementation.