added: parameterizable LeaderRank algorithm
also added visualization for this algorithm
To see the results click "Find Key vertices with leader Rank" button
You'll see the dialogue window with parameters to input:
you may input amount of vertices XOR the gap of ranks to see
after that TopRankedVertices will be coloured with red
the idea of algorithn itself:
LeaderRank(graph G, damping_factor d, convergence_threshold epsilon)
Initialize the vertex ratings: each vertex v_i has a rating R(v_i) = 1 / |V|, where |V| is the total number of vertices in graph G.
Let's construct the adjacency matrix A of graph G.
Let's construct a matrix of transition probabilities P: P = A / deg, where deg is the number of outgoing edges for each vertex.
Repeat until convergence:
4.1. Calculate the new vertex ratings: R_new = (1-d) / | V| + d P R, where d is the attenuation coefficient.
4.2. Calculate the change in ratings: diff = ||R_new - R||
4.3. If diff < epsilon, interrupt the iterations.
4.4. Assign R = R_new and continue iterating.
We display the ratings of the vertices in descending order.
added: parameterizable LeaderRank algorithm also added visualization for this algorithm To see the results click "Find Key vertices with leader Rank" button You'll see the dialogue window with parameters to input: you may input amount of vertices XOR the gap of ranks to see after that TopRankedVertices will be coloured with red
the idea of algorithn itself: LeaderRank(graph G, damping_factor d, convergence_threshold epsilon)