TABLE 3.

Pseudocode for balanced graphs (algorithm 2)^a

Step	Description
1	M[0] ← scoring matrix initialized from weighted adjacency matrix
2	ϵ ← threshold for convergence
3	E ← initial error value, set to 1
4	I ← maximum number of iterations
5	for i = 1 to I do
6	while 100E > ϵ then
7	$M[i]\leftarrow M{[i-1]}^2$
8	$M[i]\leftarrow$ apply algorithm 1 on M[i]
9	for row, column in M[i] do
10	$M{[i]}_{\text{row,column}}\leftarrow M{[i]}_{\text{row,column}}+1/M{[i]}_{\text{row,column}}$
11	end
12	$M[i]\leftarrow$ apply algorithm 1 on M[i]
13	$E\leftarrow 0$
14	$b\leftarrow 0$
15	for row, column in M[i] do
16	if $M{[i]}_{\text{row,column}}$ is not 0 then
17	$b\leftarrow b+1$
18	$E\leftarrow M{[i]}_{\text{row,column}}-M{[i-1]}_{\text{row,column}}$
19	end
20	$E\leftarrow E/b$
21	end

^aPseudocode operators “while,” “for,” “if,” “is not,” “then,” “do,” and “end” are highlighted in bold for clarity.