segment.h

中文分词系统分析库

/****************************************************************************
 *
 * Copyright (c) 2000, 2001 
 *     Machine Group
 *     Software Research Lab.
 *     Institute of Computing Tech.
 *     Chinese Academy of Sciences
 *     All rights reserved.
 *
 * This file is the confidential and proprietary property of 
 * Institute of Computing Tech. and the posession or use of this file requires 
 * a written license from the author.
 * Filename: Segment.cpp
 * Abstract:
 *           interface for the CSegment class.
 * Author:   Kevin Zhang 
 *          ([e-mail])
 * Date:     2002-4-23
 *
 * Notes:  N-Shortest paths Word segmentation
 *                
 ****************************************************************************/

#if !defined(AFX_SEGMENT_H__76DAE34D_434F_4989_8F00_64F17B5E19ED__INCLUDED_)
#define AFX_SEGMENT_H__76DAE34D_434F_4989_8F00_64F17B5E19ED__INCLUDED_

#if _MSC_VER > 1000
#pragma once
#endif // _MSC_VER > 1000
#include "DynamicArray.h"
#include "..\\Utility\\Dictionary.h"
#include "SegGraph.h"//The segmentation graph
#define MAX_WORDS 650
#define MAX_SEGMENT_NUM 10
class CSegment  
{
public:
	bool GetLastWord(PWORD_RESULT pItem, char *sWordRet);
	int GetResultCount(PWORD_RESULT pItem);
	bool OptimumSegmet(int nResultCount);
	//Word Segmentation based on optimum segmentation graph
	//After unknown word recognition
	bool Segment(char *sSentence,CDictionary &dictCore,int nResultCount=10);
	CSegment();
	virtual ~CSegment();
	PWORD_RESULT *m_pWordSeg;
	int m_nSegmentCount;
	//The segmentation result
	CDynamicArray m_graphOptimum;//The optimumized segmentation graph
	CSegGraph m_graphSeg;//The segmentation graph
protected:
	bool IsYearTime(char *sNum);
	bool GenerateWord(int ** nSegRoute, int nIndex);
};
#endif // !defined(AFX_SEGMENT_H__76DAE34D_434F_4989_8F00_64F17B5E19ED__INCLUDED_)