xiaoqichuli.h

利用贝叶斯分类原理实现多义词的消歧。首先利用训练语料进行训练

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

#if !defined(AFX_XIAOQICHULI_H__18F5F1DB_0CE4_4BBF_A72B_9225AE4E5308__INCLUDED_)
#define AFX_XIAOQICHULI_H__18F5F1DB_0CE4_4BBF_A72B_9225AE4E5308__INCLUDED_

#define STL_USING_ALL
#include "Stl.h"

#if _MSC_VER > 1000
#pragma once
#endif // _MSC_VER > 1000
/////////////////////////////////////////////////////////////////////////////
class wordID
{
public:
	string word;
	int    IDw;
	friend operator <(const wordID & left,const wordID & right)
	{
		return left.word < right.word;
	}
};

class wordPvs
{
public:
	int     IDw;
	float   Pvs;
	friend operator <(const wordPvs & left,const wordPvs & right)
	{
		return left.IDw < right.IDw;
	}
};
class wordPsk
{
public:
	int    IDw;
//	int    IDs;
	string IDs;
	float  Psk;
	float  P_LAM;
	long   begin;
	long   end;
//	vector<wordPvs>  vectWordPvsTmp;  //临时的词汇Pvs表
	friend operator <(const wordPsk & left,const wordPsk & right)
	{
		return left.IDw < right.IDw;
	}

};

class word_Tmp
{
public:
	string word;            //每句待处理的词汇
//	int    IDs;             //待输出的词汇向量
	string IDs;
};

typedef vector<wordPsk>::iterator PP;            /////////@@@@@@@@@@@@@@@@@@@@@@@@@@@@***********###########!!!!!!!!!!!


/////////////////////////////////////////////////////////////////////////////
class xiaoqichuli  
{
public:
	void CountSore(string &);
	void ReadPvs(vector<wordPvs> vectPvs, vector<wordPsk> ::iterator pIt);
	vector<wordPvs> _ReadPvs(vector<wordPsk> ::iterator pIt);
	void CountPvs(int i, vector<wordPvs> vectWordPvsTmp, vector<wordPsk> ::iterator pIt);
//	void SearchPsk();
	void SearchIDw();
	void CreatePskList();
	void CreateWordIDList();
	void ReadCorpus(string, char *);
	xiaoqichuli();
	virtual ~xiaoqichuli();
private:
	vector<wordID>   vectWordID;      //词汇ID表
	vector<wordPsk>  vectWordPsk;     //词汇Psk表
//	vector<string>   vectWordTmp;     
	vector<int>      vectIDTmp;       //每句临时的词汇ID
	vector<word_Tmp> vectWordTmp;     //每句待处理的词汇
//	vector<wordIDs>  vectWordIDs;  
	double SoreTmp;
	vector<pair<double, string> > vectSore;    //只考虑发射概率
	vector<pair<double, string> > vectSore_T;  //条件概率和发射概率的乘积


};

#endif // !defined(AFX_XIAOQICHULI_H__18F5F1DB_0CE4_4BBF_A72B_9225AE4E5308__INCLUDED_)