stanford 词性标注 性能慢

[zbl430]

代码如下

#! /usr/bin/env python
# -*- coding: utf-8 -*-

"""
 *
"""
import sys
import time
from nltk.tag import StanfordPOSTagger

reload(sys)
sys.setdefaultencoding('utf-8')

model_filename = "./data/modles/pos.tagger"
path_to_jar = "./stanford-postagger.jar"
Tagger = StanfordPOSTagger(model_filename=model_filename, path_to_jar=path_to_jar)

if __name__ == "__main__":
    st = time.time()
    print Tagger.tag([u"你的", u"百度", u"打人"]), time.time()-st
    print Tagger.tag([u"你的", u"百度", u"打人"]), time.time()-st
    print Tagger.tag([u"你的", u"百度", u"打人"]), time.time()-st

输出:

[(u'\u4f60\u7684', u'nz'), (u'\u767e\u5ea6', u'nz'), (u'\u6253\u4eba', u'v')] 5.10674095154 s
[(u'\u4f60\u7684', u'nz'), (u'\u767e\u5ea6', u'nz'), (u'\u6253\u4eba', u'v')] 10.2533240318 s
[(u'\u4f60\u7684', u'nz'), (u'\u767e\u5ea6', u'nz'), (u'\u6253\u4eba', u'v')] 16.8123478889 s

速度竟然如此慢，请大佬赐教，是我打开方式不对还是怎么？？

[holajamc]

既然已经用了 stanford 不如直接用 hankcs

[knightdf]

别用 python 调了，好像每次起一个 jvm 做的，当然慢了