gistfile1.py

#!/usr/bin/env python
# HITS (on memory)
# using co-citation matrix

from numpy import *
from numpy.linalg import *

class HITS():
    def __init__(self, adj_matrix):
        self._adj = adj_matrix

    def calc(self):
        cit_matrix = dot(self._adj.T, self._adj)
        bib_matrix = dot(self._adj, self._adj.T)
        e_val,e_vec = eig( cit_matrix )
        auth = sorted( zip( e_val, e_vec.T ), reverse=True )
        e_val,e_vec = eig( bib_matrix )
        hub  = sorted( zip( e_val, e_vec.T ), reverse=True )
        return (auth[0][1] / sum(auth[0][1]),
                hub[0][1] / sum(hub[0][1]))


Adj = array( ((0, 0, 1, 1),
              (0, 0, 1, 1),
              (1, 1, 0, 0),
              (0, 1, 1, 0)) )

hits = HITS(Adj)
(auth,hub) = hits.calc()
print "auth:", auth
print "hub:", hub