kmaehashi · June 20, 2016 08:50
diff --git a/README.md b/README.md
diff --git a/benchmark.py b/benchmark.py
 # -*- coding: utf-8 -*-

 import jubatus
 import time

 # create labeled datum
 sentence = 'あいうえおかきくけこさしすせそたちつてとなにぬねのはひふへほabcd'  # 100 bytes
 sentences = {}
 for i in range(100):
  sentences['data_{0}'.format(i)] = sentence
 d = jubatus.classifier.types.LabeledDatum('pos', jubatus.common.Datum(sentences))

 # create classifier
 c = jubatus.Classifier("127.0.0.1", 9199, '', 0)

 total = 0
 count = 0
 while True:
  batch_size = 1000
  begin = time.time()
  for i in range(batch_size):
    c.train([d])
  total += time.time() - begin
  count += batch_size
  print('[{0} queries sent] ... average latency: {1} ms'.format(count, (total * 1000 / count) ))
diff --git a/config.json.bin-bin b/config.json.bin-bin
 {
  "method": "perceptron",
  "converter": {
    "string_filter_types": {},
    "string_filter_rules": [],
    "num_filter_types": {},
    "num_filter_rules": [],
    "string_types": {},
    "string_rules": [
      { "key": "*", "type": "space", "sample_weight": "bin", "global_weight": "bin"}
    ],
    "num_types": {},
    "num_rules": []
  }
 }
diff --git a/config.json.tf-bm25 b/config.json.tf-bm25
 {
  "method": "perceptron",
  "converter": {
    "string_filter_types": {},
    "string_filter_rules": [],
    "num_filter_types": {},
    "num_filter_rules": [],
    "string_types": {},
    "string_rules": [
      { "key": "*", "type": "space", "sample_weight": "tf", "global_weight": "bm25"}
    ],
    "num_types": {},
    "num_rules": []
  }
 }
diff --git a/config.json.tf-idf b/config.json.tf-idf
 {
  "method": "perceptron",
  "converter": {
    "string_filter_types": {},
    "string_filter_rules": [],
    "num_filter_types": {},
    "num_filter_rules": [],
    "string_types": {},
    "string_rules": [
      { "key": "*", "type": "space", "sample_weight": "tf", "global_weight": "idf"}
    ],
    "num_types": {},
    "num_rules": []
  }
 }
diff --git a/result.md b/result.md
	# -- coding: utf-8 --

	import jubatus
	import time

	# create labeled datum
	sentence = 'あいうえおかきくけこさしすせそたちつてとなにぬねのはひふへほabcd' # 100 bytes
	sentences = {}
	for i in range(100):
	sentences['data_{0}'.format(i)] = sentence
	d = jubatus.classifier.types.LabeledDatum('pos', jubatus.common.Datum(sentences))

	# create classifier
	c = jubatus.Classifier("127.0.0.1", 9199, '', 0)

	total = 0
	count = 0
	while True:
	batch_size = 1000
	begin = time.time()
	for i in range(batch_size):
	c.train([d])
	total += time.time() - begin
	count += batch_size
	print('[{0} queries sent] ... average latency: {1} ms'.format(count, (total * 1000 / count) ))
	{
	"method": "perceptron",
	"converter": {
	"string_filter_types": {},
	"string_filter_rules": [],
	"num_filter_types": {},
	"num_filter_rules": [],
	"string_types": {},
	"string_rules": [
	{ "key": "*", "type": "space", "sample_weight": "bin", "global_weight": "bin"}
	],
	"num_types": {},
	"num_rules": []
	}
	}