Algorithm 1. Spectral Clustering
Input: Connectivity matrix W with size n×n, and the threshold T of Ncut for partitioning.
Output: Clusters of nodes.
1.	Compute the normalized Laplacian L of W.
2.	Solve eigenvectors of L with the smallest eigenvalues.
3.	Use the eigenvector with the second smallest eigenvalue to bi-partition the graph, and then compute the corresponding Ncut value.
4.	If Ncut < T, bi-partition the graph, and repeat the algorithm on two bi-partitioned sub-graphs.
5.	Else Return without bi-partitioning the graph.