Calculate pagerank of twitter follow network.

tw_pagerank.rb

#!/usr/bin/env ruby
$:.unshift(File.dirname(__FILE__))
require 'twitter'
require 'mysql2-cs-bind'
require 'yaml'
require 'pagerank'

# parameter
DUMPING_FACTOR = 0.85

# write twitter token and database settings in config.yml
config = YAML::load_file("path/to/config.yml")
tw = Twitter::Client.new(config['twitter'])
client = Mysql2::Client.new(config['database'])

# twitter account name
user_name = "hongo35"

# get user ids which "user_name" is following
friend_ids = []
init = []
friend_cursor = -1
while friend_cursor != 0
  friends = tw.friend_ids(user_name, :cursor => friend_cursor)
  friends.ids.each do |id|
    friend_ids.push(id)
    init.push(1.0)
  end

  friend_cursor = friends.next_cursor
  sleep(60)
end

# get foaf(friends of a friend) ids
friend_ids.each do |id|
  begin
    id_cursor = -1
    while id_cursor != 0
      foaf = tw.friend_ids(id, :cursor => id_cursor)
      foaf.ids.each do |friend_id|
        client.xquery("insert into links(user_id, friend_id) values(?,?)", id, friend_id)
      end 

      id_cursor = foaf.next_cursor
      sleep(60)
    end 
  rescue
  end 
end

# friends follow network matrix
matrix_hash = {}
matrix_size = friend_ids.size
friend_ids.each do |row_id|
  matrix_hash[row_id] = {}
  friend_ids.each do |column_id|
    if row_id == column_id
      matrix_hash[row_id][column_id] = 1 
    else
      matrix_hash[row_id][column_id] = 0
    end
  end
end

res = client.xquery("select * from links where friend_id in (#{friend_ids.join(',')})")
res.each do |r|
  matrix_hash[r['user_id']][r['friend_id']] = 1
end

matrix = []
matrix_hash.each do |key, value|
  matrix_tmp = []
  value.each do |k,v|
    matrix_tmp.push(v)
  end
  matrix.push(matrix_tmp)
end

# pagerank initial array
init.each_with_index do |val, i|
  init[i] = val / matrix_size
end

# calculate pagerank
pr = PageRank.new(matrix)
rank = pr.calc(init, DUMPING_FACTOR)
rank.each_with_index do |r, i|
  puts "#{friend_ids[i]},#{r}"
end