說到分詞大家肯定一般認為是很高深的技術,但是今天作者用短短几十行程式碼就搞定了,感嘆python很強大啊!作者也很強大。不過這個只是正向最大匹配,沒有機器學習能力
注意:使用前先要下載搜狗詞庫
# -*- coding:utf-8 -*- #寫了一個簡單的支援中文的正向最大匹配的機械分詞,其它不用解釋了,就幾十行程式碼 #附:搜狗詞庫下載地址:http://vdisk.weibo.com/s/7RlE5 import string __dict = {} def load_dict(dict_file='words.dic'): #載入詞庫,把詞庫載入成一個key為首字元,value為相關詞的列表的字典 words = [unicode(line, 'utf-8').split() for line in open(dict_file)] for word_len, word in words: first_char = word[0] __dict.setdefault(first_char, []) __dict[first_char].append(word) #按詞的長度倒序排列 for first_char, words in __dict.items(): __dict[first_char] = sorted(words, key=lambda x:len(x), reverse=True) def __match_ascii(i, input): #返回連續的英文字母,數字,符號 result = '' for i in range(i, len(input)): if not input[i] in string.ascii_letters: break result += input[i] return result def __match_word(first_char, i , input): #根據當前位置進行分詞,ascii的直接讀取連續字元,中文的讀取詞庫 if not __dict.has_key(first_char): if first_char in string.ascii_letters: return __match_ascii(i, input) return first_char words = __dict[first_char] for word in words: if input[i:i+len(word)] == word: return word return first_char def tokenize(input): #對input進行分詞,input必須是uncode編碼 if not input: return [] tokens = [] i = 0 while i < len(input): first_char = input[i] matched_word = __match_word(first_char, i, input) tokens.append(matched_word) i += len(matched_word) return tokens if __name__ == '__main__': def get_test_text(): import urllib2 url = "http://news.baidu.com/n?cmd=4&class=rolling&pn=1&from=tab&sub=0" text = urllib2.urlopen(url).read() return unicode(text, 'gbk') def load_dict_test(): load_dict() for first_char, words in __dict.items(): print '%s:%s' % (first_char, ' '.join(words)) def tokenize_test(text): load_dict() tokens = tokenize(text) for token in tokens: print token tokenize_test(unicode(u'美麗的花園裡有各種各樣的小動物')) tokenize_test(get_test_text())我也學習啦~~~