#?-*-?coding:?cp936?-*-
“““
基于支持向量機的郵件分類系統
使用01標記詞的出現與否
“““
import?re
from?math?import?*
from?SVMKernel?import?*
#切割文本統計詞頻
def?splitText（bigString）:
???
????wordlist={}
????rtnList=[]
????wordFreqList={}
????#分詞
????listofTokens=re.split（r‘\W*‘bigString）
????length=len（listofTokens）
????for?token?in?listofTokens:
????????if??not?wordlist.has_key（token）:
????????????wordlist[token]=1
????????????rtnList.append（token）
????????else:
????????????wordlist[token]+=1
????????wordFreqList[token]=float（wordlist[token]）/length
????return?rtnListwordFreqList
???????
#統計單詞反文檔頻率
def?docFre（word）:
????fre=0
????for?i?in?range（126）:
????????if?word?in?re.split（r‘\W*‘open（‘spam/%d.txt‘?%?i）.read（））:
?????????????fre+=1
???
????return?float（fre）/25

????

#特征詞提取，這

?屬性????????????大小?????日期????時間???名稱
-----------?---------??----------?-----??----

?????文件????????148??2010-10-23?17:11??email\ham\1.txt

?????文件?????????86??2010-10-23?17:13??email\ham\10.txt

?????文件????????130??2010-10-23?17:13??email\ham\11.txt

?????文件????????182??2010-10-23?09:16??email\ham\12.txt

?????文件????????174??2010-10-23?17:13??email\ham\13.txt

?????文件????????172??2010-10-23?17:13??email\ham\14.txt

?????文件????????531??2010-10-23?09:21??email\ham\15.txt

?????文件?????????90??2010-10-23?09:21??email\ham\16.txt

?????文件????????464??2010-10-23?09:22??email\ham\17.txt

?????文件????????175??2010-10-23?09:23??email\ham\18.txt

?????文件????????161??2010-10-23?17:14??email\ham\19.txt

?????文件????????234??2010-10-23?08:48??email\ham\2.txt

?????文件????????208??2010-10-23?09:26??email\ham\20.txt

?????文件????????234??2010-10-23?09:27??email\ham\21.txt

?????文件????????330??2010-10-23?09:28??email\ham\22.txt

?????文件????????608??2010-10-23?17:15??email\ham\23.txt

?????文件?????????42??2010-10-23?09:33??email\ham\24.txt

?????文件?????????89??2010-10-23?09:34??email\ham\25.txt

?????文件????????371??2010-10-23?08:49??email\ham\3.txt

?????文件????????207??2010-10-23?08:50??email\ham\4.txt

?????文件????????114??2010-10-23?17:11??email\ham\5.txt

?????文件???????1464??2010-10-23?17:12??email\ham\6.txt

?????文件????????109??2010-10-23?17:12??email\ham\7.txt

?????文件????????638??2010-10-23?08:58??email\ham\8.txt

?????文件????????146??2010-10-23?09:01??email\ham\9.txt

?????文件????????238??2010-10-23?08:28??email\spam\1.txt

?????文件????????217??2010-10-23?08:36??email\spam\10.txt

?????文件????????414??2010-10-23?08:37??email\spam\11.txt

?????文件????????188??2010-10-23?08:37??email\spam\12.txt

?????文件????????252??2010-10-23?08:38??email\spam\13.txt

............此處省略30個文件信息