解壓縮後取出以下檔案:
訓練資料:icwb2-data/training/pku_ training.utf8
訓練資料:icwb2-data/testing/pku_ test.utf8
測試資料:icwb2-data/testing/pku_ test.utf8正確分詞結果:wendboldb. test_ gold.utf8評分工具:icwb2-data/script/socre2 演算法描述演算法是最簡單的正向最大匹配(FMM):用訓練資料產生一個字典對測試資料對測試資料左到右掃描,遇到一個最長的詞,就切分下來,直到句子結束註:這是最初的算法,這樣做代碼可以控制在60行內,後來看測試結果發現沒有很好地處理數字問題, 才又增加了對數字的處理。 3 源代碼及註釋#! /usr/bin/env python # -*- coding: utf-8 -*- # Author: minix # Date: 2013-03-20 import codecs import sys # 由规则处理的一些特殊符号 numMath = [u'0', u'1', u'2', u'3', u'4', u'5', u'6', u'7', u'8', u'9'] numMath_suffix = [u'.', u'%', u'亿', u'万', u'千', u'百', u'十', u'个'] numCn = [u'一', u'二', u'三', u'四', u'五', u'六', u'七', u'八', u'九', u'〇', u'零'] numCn_suffix_date = [u'年', u'月', u'日'] numCn_suffix_unit = [u'亿', u'万', u'千', u'百', u'十', u'个'] special_char = [u'(', u')'] def proc_num_math(line, start): """ 处理句子中出现的数学符号 """ oldstart = start while line[start] in numMath or line[start] in numMath_suffix: start = start + 1 if line[start] in numCn_suffix_date: start = start + 1 return start - oldstart def proc_num_cn(line, start): """ 处理句子中出现的中文数字 """ oldstart = start while line[start] in numCn or line[start] in numCn_suffix_unit: start = start + 1 if line[start] in numCn_suffix_date: start = start + 1 return start - oldstart def rules(line, start): """ 处理特殊规则 """ if line[start] in numMath: return proc_num_math(line, start) elif line[start] in numCn: return proc_num_cn(line, start) def genDict(path): """ 获取词典 """ f = codecs.open(path,'r','utf-8') contents = f.read() contents = contents.replace(u'\r', u'') contents = contents.replace(u'\n', u'') # 将文件内容按空格分开 mydict = contents.split(u' ') # 去除词典List中的重复 newdict = list(set(mydict)) newdict.remove(u'') # 建立词典 # key为词首字,value为以此字开始的词构成的List truedict = {} for item in newdict: if len(item)>0 and item[0] in truedict: value = truedict[item[0]] value.append(item) truedict[item[0]] = value else: truedict[item[0]] = [item] return truedict def print_unicode_list(uni_list): for item in uni_list: print item, def divideWords(mydict, sentence): """ 根据词典对句子进行分词, 使用正向匹配的算法,从左到右扫描,遇到最长的词, 就将它切下来,直到句子被分割完闭 """ ruleChar = [] ruleChar.extend(numCn) ruleChar.extend(numMath) result = [] start = 0 senlen = len(sentence) while start < senlen: curword = sentence[start] maxlen = 1 # 首先查看是否可以匹配特殊规则 if curword in numCn or curword in numMath: maxlen = rules(sentence, start) # 寻找以当前字开头的最长词 if curword in mydict: words = mydict[curword] for item in words: itemlen = len(item) if sentence[start:start+itemlen] == item and itemlen > maxlen: maxlen = itemlen result.append(sentence[start:start+maxlen]) start = start + maxlen return result def main(): args = sys.argv[1:] if len(args) < 3: print 'Usage: python dw.py dict_path test_path result_path' exit(-1) dict_path = args[0] test_path = args[1] result_path = args[2] dicts = genDict(dict_path) fr = codecs.open(test_path,'r','utf-8') test = fr.read() result = divideWords(dicts,test) fr.close() fw = codecs.open(result_path,'w','utf-8') for item in result: fw.write(item + ' ') fw.close() if __name__ == "__main__": main()
使用tail 查看結果文件最後幾行的總體評分,另外socre.utf8中還提供了大量的比較結果, 可以用於發現自己的分詞結果在哪裡做的不夠好注:整個測試過程都在Ubuntu下完成$ python dw.py pku_training.utf8 pku_test.utf8 pku_result.utf8$ perl score pku_training.utf8 pku_test_gold.utf8 pkuresult.f588 INSERTIONS: 0 DELETIONS: 0SUBSTITUTIONS: 0NCHANGE: 0NTRUTH: 27UEUE777% 1.000TEST WORDS PRECISION: 1.000=== SUMMARY:== = TOTAL INSERTIONS: 4623=== TOTAL DELETIONS: 1740=== TOTAL SUBSTITUTIONS: TOTAL TRUE WORD COUNT: 104372=== TOTAL TEST WORD COUNT: 107255=== TOTAL TRUE WORDS RECALL: 0.920=== TOTAL TEST WORDS PRECISION: 0.8955
=== TOTAL TEST WORDS PRECISION: 0.8955% 0.940
=== OOV Recall Rate: 0.917
=== IV Recall Rate: 0.966
Rate: 0.966基於詞典的FMM演算法是非常基礎的分詞演算法,效果沒那麼好,但不過簡單,也不能我可能還會用Python實現其它的分詞演算法。另外一個感受是,看書的時候盡量多去實現,這樣會讓你有足夠的熱情去關注理論的每一個細節,不會感到那麼枯燥無力。