ZH奶酪:自然语言处理工具LTP语言云调用方法

前言

LTP语言云平台

      不支持离线调用;

  支持分词、词性标注、命名实体识别、依存句法分析、语义角色标注

  不支持自定义词表,但是你可以先用其他支持自定义分词的工具(例如中科院的NLPIR)把文本进行分词,再让ltp帮你标注

     支持C#、Go、Java、JavaScript、Nodejs、PHP、Python、R、Ruby等语言调用;  

     还有一些错误响应频率限制重要说明(这几个我至今也没用到);

正文

官方网址:http://www.ltp-cloud.com/

使用文档:http://www.ltp-cloud.com/document/

在线演示:http://www.ltp-cloud.com/demo/

各种语言调用实例可以到Github上下载:https://github.com/HIT-SCIR/ltp-cloud-api-tutorial

例如Python版本的:https://github.com/HIT-SCIR/ltp-cloud-api-tutorial/tree/master/Python

Step1:注册

这个网址申请一个API key,稍后会用到;

Step2:一个简单的例子(Python版)

(1)复制代码:从Github上复制一段代码(取决于你使用的语言和所需的功能)

(2)修改代码:

  <1>把 api_key = "YourApiKey" 中的 "YourApiKey" 修改成你Step1申请的API Key;

  <2>把 text = "我爱北京天安门" 修改成你要处理的文本;

  <3>根据需求设置不同的参数(其实只需要api_key,text,pattern,format四个参数就够了,仔细看下pattern):

   

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

# This example shows how to use Python to access the LTP API to perform full
# stack Chinese text analysis including word segmentation, POS tagging, dep-
# endency parsing, name entity recognization and semantic role labeling and
# get the result in specified format.

import urllib2, urllib
import sys

if __name__ == '__main__':
    if len(sys.argv) < 2 or sys.argv[1] not in ["xml", "json", "conll"]:
        print >> sys.stderr, "usage: %s [xml/json/conll]" % sys.argv[0]
        sys.exit(1)

    uri_base = "http://ltpapi.voicecloud.cn/analysis/?"
    api_key  = "YourApiKey"
    text     = "我爱北京天安门"
    # Note that if your text contain special characters such as linefeed or '&',
    # you need to use urlencode to encode your data
    text     = urllib.quote(text)
    format   = sys.argv[1]
    pattern  = "all"

    url      = (uri_base
               + "api_key=" + api_key + "&"
               + "text="    + text    + "&"
               + "format="  + format  + "&"
               + "pattern=" + "all")

    try:
        response = urllib2.urlopen(url)
        content  = response.read().strip()
        print content
    except urllib2.HTTPError, e:
        print >> sys.stderr, e.reason

Step3:运行

如果要批量处理txt或者xml文件,需要自己写一段批量处理的代码,下边是我之前项目中用到的一段批量处理某一目录下txt文件代码(就是加了一层循环和设置了一个输出):

 1 # -*- coding: utf-8 -*-
 2 #!/usr/bin/env python
 3 
 4 # This example shows how to use Python to access the LTP API to perform full
 5 # stack Chinese text analysis including word segmentation, POS tagging, dep-
 6 # endency parsing, name entity recognization and semantic role labeling and
 7 # get the result in specified format.
 8 
 9 import urllib2, urllib
10 import sys
11 
12 if __name__ == '__main__':
13     uri_base = "http://ltpapi.voicecloud.cn/analysis/?"
14     api_key  = "7132G4z1HE3S********DSxtNcmA1jScSE5XumAI"
15 
16     f = open("E:\\PyProj\\Others\\rite_sentence.txt")
17     fw = open("E:\\PyProj\\Others\\rite_pos.txt",'w')
18 
19     line = f.readline()
20     while(line):
21         text     = line
22         # Note that if your text contain special characters such as linefeed or '&',
23         # you need to use urlencode to encode your data
24         text     = urllib.quote(text)
25         format   = "plain"
26         pattern  = "pos"
27 
28         url      = (uri_base
29                    + "api_key=" + api_key + "&"
30                    + "text="    + text    + "&"
31                    + "format="  + format  + "&"
32                    + "pattern=" + pattern)
33 
34         try:
35             response = urllib2.urlopen(url)
36             content  = response.read().strip()
37             print content
38             fw.write(line+content+'\n')
39         except urllib2.HTTPError, e:
40             print >> sys.stderr, e.reason
41         line = f.readline()
42     fw.close()
43     f.close()

 

posted @ 2015-06-18 10:28  ZH奶酪  阅读(3796)  评论(9编辑  收藏  举报