extractor.h

中文邮件过滤。对训练邮件分词训练贝叶斯模型。然后对测试邮件分类

#define HAMDIR  "samples\\ham\\"
#define SPAMDIR "samples\\spam\\"

typedef struct Feature
{
	char word[18];
	int tf;
    Feature *next;
}Feature;

typedef struct dict
{
	char str[18];
	int len;
	dict *next;
}dict;

typedef struct Document
{
	int ID;
	int c;
	Feature *feature[4000];
}Document;

/*
typedef Feature ;
typedef  dict ;
typedef Document ;
*/
void readDict();
int getIndex(char wordh,char wordl);
void statisticHam();
void statisticSpam();
void phaseSta();