seggraph.h

计算所汉语词法分析系统ICTCLAS介绍 词是最小的能够独立活动的有意义的语言成分。 但汉语是以字为基本的书写单位

/****************************************************************************
 *
 * 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:  SegGraph.h
 * Abstract:
 *           interface for the Word Segmentation Directed Graph.
 * Author:   Kevin Zhang 
 *          ([e-mail])
 * Date:     2002-4-18
 *
 * Notes:
 *                
 * 
 ****************************************************************************/
// SegGraph.h: interface for the CSegGraph class.
//
//////////////////////////////////////////////////////////////////////

#if !defined(AFX_SEGGRAPH_H__E051AD07_F18C_4171_94D6_39742104EB62__INCLUDED_)
#define AFX_SEGGRAPH_H__E051AD07_F18C_4171_94D6_39742104EB62__INCLUDED_

#if _MSC_VER > 1000
#pragma once
#endif // _MSC_VER > 1000
#include "..\\Utility\\Dictionary.h"//Dictionary
#include "DynamicArray.h"//Dynamic array
#define MAX_FREQUENCE    2079997//1993123+86874
#define MAX_SENTENCE_LEN 2000

class CSegGraph  
{
public:
	bool GenerateWordArray(char *sSentence,CDictionary &dictCore);
	bool GenerateWordNet(char *sSentence,CDictionary &dictCore);
	//Generate the segmentation word net according the original sentence
	//sSentence: the sentence
	//dictCore: core dictionary
	CSegGraph();
	virtual ~CSegGraph();
	char m_sAtom[MAX_SENTENCE_LEN][WORD_MAXLENGTH];//pAtoms: the buffer for returned segmented atoms
						// Such as a Chinese Char, digit, single byte, or delimiters
	int m_nAtomLength[MAX_SENTENCE_LEN];//Save the individual length of atom in the array
	int m_nAtomPOS[MAX_SENTENCE_LEN];//pAtoms: the POS property 
	unsigned int m_nAtomCount;//The count of atoms

	CDynamicArray m_segGraph;	//segGraph: The segmentation word graph
protected:
	bool AtomSegment(char *sSentence);
	//Segment the atomic members from the original sentence
	//sSentence: the sentence
	//pAtoms: the buffer for returned segmented atoms
	//        Such as a Chinese Char, digit, single byte, or delimiters
};

#endif // !defined(AFX_SEGGRAPH_H__E051AD07_F18C_4171_94D6_39742104EB62__INCLUDED_)