Jubatus classifier test

jubatus.rb

# -*- coding: utf-8 -*-

$host = "localhost"
$port = 9199
$name = "paper_genre"

require 'json'
require 'find'

require 'jubatus/classifier/client'
require 'jubatus/classifier/types'

def train(client)
  # prepare training data
  # predict the last ones (that are commented out)
  
  chem_files = []
  Find.find('chem/'){|f|
    chem_files.push(f) if File.file? f
  }
  bio_files = []
  Find.find('bio/'){|f|
    bio_files.push(f) if File.file? f
  }
  # p chem_files
  # p bio_files
  chem_data = []
  chem_files.each{|f|
    abs = IO.readlines(f).join('\n')
    chem_data += [["chem",Jubatus::Classifier::Datum.new([["abstract",abs]],[])]]
  }
  bio_data = []
  bio_files.each{|f|
    abs = IO.readlines(f).join('\n')
    bio_data += [["bio",Jubatus::Classifier::Datum.new([["abstract",abs]],[])]]
  }

  train_data = chem_data + bio_data

  # p train_data

  # training data must be shuffled on online learning!
  train_data.sort_by{rand}

  # run train
  r = client.train($name, train_data)
  p r
end

def predict(client)
  # predict the last shogun
  data =
    [
     Jubatus::Classifier::Datum.new([["abstract", "Asprin was synthesized."]], []),
     Jubatus::Classifier::Datum.new([["abstract", "This DNA sequence is specific to E coli."]], []),
     Jubatus::Classifier::Datum.new([["abstract", "Toxicity of carbon nanotube is still unclear."]], []),
     Jubatus::Classifier::Datum.new([["abstract", "Proteinogenic amino acids are amino acids that are precursors to proteins."]], []),
    ]
  res = client.classify($name, data)
  p res
  data.each { |d|
    res = client.classify($name, [d])
    # get the predicted genre
    # p res
    puts res[0].max{ |x, y| x[1] <=> y[1]}[0] + ": " + d.string_values[0][1]
  }
end

# connect to the jubatus
client = Jubatus::Classifier::Client::Classifier.new($host, $port)
# run example
train(client)
predict(client)