textclassify.h

用来文本分类的

// TextClassify.h: interface for the TextClassify class.
//
//////////////////////////////////////////////////////////////////////

#include "ClassTree.h"
#include "wlistname.h"
class TextClassify  
{
public:
	TextClassify(char *file,char *strStopWordFile);
	virtual ~TextClassify();
	bool InitialTree();
	bool UpdateTree(char *strUpdatePath);
	bool SaveTree(char * SaveTreePath);
	bool TrainTree(CString strTrainDirectory,int WordFreq,float Weight,float Central);
	bool Classify(char * strFileName,CString& strClassifyResult);
private:
	//train part function
	void PostOrderTrain();
	void VisitNode();
	void tfStat(char *fileName,int *pTFArray);
	void CCM(int ** pkinddoct,int nNumOfKind, WORDVECTOR **pwordlist);
	//classify part function
	int ClassifyKind(int *pTFArray,int nNumOfKind,WORDVECTOR **pWordVectorArray);//return the kind
//class member
	ClassTree tree;
	WListName wlisti;
public:
	int WordFreqThreshold;
	float WeightThreshold;
	float CentralThreshold;
	CString strTrainPath;
	int nNumOfStopWord;
	char (*listStopWord)[10];
	//test add by leiyun
	void veracity(double *veracity,CString m_strResultPath);
	void clear(CString m_strResultPath);
	void test(double weight,double central,int frequence);
	//test end


};