browsedir.cpp

贝叶斯学习算法源码

				swap[m].UnNormalMail_Prior = x[i].UnNormalMail_Prior;
				i ++;
			}
			else
			{
				//Copy(swap[m],x[j]);
				strcpy(swap[m].str,x[j].str);
				swap[m].gain = x[j].gain;
				swap[m].NormalMail = x[j].NormalMail;			
				swap[m].NormalMail_Prior = x[j].NormalMail_Prior;
				swap[m].UnNormalMail = x[j].UnNormalMail;
				swap[m].UnNormalMail_Prior = x[j].UnNormalMail_Prior;
				 j++;
			}
		}

		while(i <=  u1)
		{
			//Copy(swap[m],x[i]);
			strcpy(swap[m].str,x[i].str);
			swap[m].gain = x[i].gain;
			swap[m].NormalMail = x[i].NormalMail;			
			swap[m].NormalMail_Prior = x[i].NormalMail_Prior;
			swap[m].UnNormalMail = x[i].UnNormalMail;
			swap[m].UnNormalMail_Prior = x[i].UnNormalMail_Prior;
			m ++;
			i ++;
		}
		while(j <= u2)
		{
			//Copy(swap[m],x[j]);
			strcpy(swap[m].str,x[j].str);
			swap[m].gain = x[j].gain;
			swap[m].NormalMail = x[j].NormalMail;			
			swap[m].NormalMail_Prior = x[j].NormalMail_Prior;
			swap[m].UnNormalMail = x[j].UnNormalMail;
			swap[m].UnNormalMail_Prior = x[j].UnNormalMail_Prior;
			m ++;
			j ++;
		}
		l1 = u2 + 1;	
	}
	for( i =l1;i < n; i ++,m++)
	{
		//Copy(swap[m],x[i]);
		strcpy(swap[m].str,x[i].str);
		swap[m].gain = x[i].gain;
		swap[m].NormalMail = x[i].NormalMail;			
		swap[m].NormalMail_Prior = x[i].NormalMail_Prior;
		swap[m].UnNormalMail = x[i].UnNormalMail;
		swap[m].UnNormalMail_Prior = x[i].UnNormalMail_Prior;
	}
	return 0;

}

int CBrowseDir::MergesortInString(Feature feature[], int Num)
{
	Feature *swap = new Feature[Num];
	if(swap == 0)
		return -1;
	int k = 1,i;
	
	while( k < Num)
	{
		MergeInString(feature,swap,k,Num);
		//cout << " k = " << k << endl;
		for( i = 0;i < Num ;i ++)
		{
			strcpy(feature[i].str,swap[i].str);
			feature[i].gain = swap[i].gain;
			feature[i].NormalMail = swap[i].NormalMail;			
			feature[i].NormalMail_Prior = swap[i].NormalMail_Prior;
			feature[i].UnNormalMail = swap[i].UnNormalMail;
			feature[i].UnNormalMail_Prior = swap[i].UnNormalMail_Prior;
		}
		//cout << endl;
		 k = k * 2;
	}

	delete swap;
	return 0;

}

int CBrowseDir::ReadStopList(const char *stoptext)
{
	char str[40];
	memset(str,0,40);
	int size = 500;
	stopwordArray = new StopWord[320];
	memset(stopwordArray[0].str,0,40);
	int eof,Len;
	int i = 1;

	FILE *readstop;
	readstop = fopen(stoptext,"r"); 
	
	if(readstop == 0)
		return -1;

	for(eof=fscanf(readstop,"%s",&str);eof!=EOF && eof>0;eof=fscanf(readstop,"%s",&str))
	{
		Len = strlen(str);
		str[Len] = 0;
		if(Len != 0)
		{
			if(i >= 320)
			{
				cout << "StopWord  array out run !";
				return -1;
			}
			memset(stopwordArray[i].str,0,40);
			strcpy(stopwordArray[i].str,str);
			i ++ ;
		}
		memset(str,0,40);
	} 
	fclose(readstop);
	return i - 1;
}

bool CBrowseDir::Del_StopWord(const char *Word, int Num)
{
	bool flag = false;

	if(strlen(Word) < 3 )   
	{
		flag = true;
		return flag;
	}

	int low = 1,high = Num;
	int mid = 0;
	int temp;

	while(low <= high)        //快速查找法
	{
		mid = (low + high) / 2;
		temp = strcmp (Word,stopwordArray[mid].str);
		if(temp == 0)
		{
			flag = true;
			return flag;
		}
		else if(temp < 0)
			high = mid - 1;
		else 
			low = mid + 1;
	}

	return flag;
}

int CBrowseDir::CalculateGain(const char *str ,Feature feature[], int Num,const char *Output)
{
	if(str == 0)
		return -1;

	ifstream inFile(str,ios::in);          //读取信息KeyInfo.dat文件，根据它读取词库
	if(!inFile)
	{
		cerr << "Input File could not be opened."  << endl;
		exit(1);
	}

	WordReservoir Info[2],temp; //建立词库信息数组
	
	while(inFile >> temp.ClassFielName >> temp.ClassFlag >> temp.FileSum >> temp.WordsSum)
	{
		switch(temp.ClassFlag)
		{
		case 0:     //0 表示垃圾邮件构成的词库
			strcpy(Info[0].ClassFielName,temp.ClassFielName);
			Info[0].ClassFlag = temp.ClassFlag;
			Info[0].FileSum = temp.FileSum;
			Info[0].WordsSum = temp.WordsSum;
			break;
		case 1:    //1表示正常邮件构成的词库
			strcpy(Info[1].ClassFielName,temp.ClassFielName);
			Info[1].ClassFlag = temp.ClassFlag;
			Info[1].FileSum = temp.FileSum;
			Info[1].WordsSum = temp.WordsSum;
		default:
			break;
		}
	}

	//记录正常邮件和垃圾邮件之和double FileNumber0_1
	double FileNumber0_1 = (double)((Info[0].FileSum + Info[1].FileSum) * 1.00);   

	/************************************************************************/
	/* 打开Info[]中的文件，并将词组导到数组中来                             */
	Item Credit;
	int i = 0;
	double WordsNumber[2];        //存储每个类别的单词的总数目
	Item *M = new Item[70000];
	Item *N = new Item[70000];

	if( M == 0 )
		return -1;
	if( N == 0)
		return -1;
	
	if(Info[0].ClassFielName == 0)    //词库为空,返回(垃圾邮件为标志0)
		return -1;
	ifstream inInfoFile1(Info[0].ClassFielName,ios::in);
	if(!inInfoFile1)
	{
		cerr << "Input File could not be opened."  << endl;
		exit(1);
	}

	//读取垃圾邮件词库
	i = 0;
	while(inInfoFile1 >> Credit.str >> Credit.freq >> Credit.density 
		>> Credit.Sequence >>Credit.FileCnt >>Credit.flag)
	{
		Copy(M[i],Credit);
		M[i].flag  = Info[0].ClassFlag;
		M[i].SetInfo = 0;
		i++;
	}

	 int Record = Unique(M,i);    //计算熵之前，先排除重复的记录并由Record记录无重复的数量

	WordsNumber[0] = (double)Record;    //记录垃圾邮件得到的词汇总数

	i = 0;
	if(Info[1].ClassFielName == 0)     //词库为空,返回(正常邮件标志为1)
		return -1;

	ifstream inInfoFile2(Info[1].ClassFielName,ios::in);
	if(!inInfoFile2)                     
	{
		cerr << "Input File could not be opened."  << endl;
		exit(1);
	}

	//读取正常邮件词库
	i = 0;
	while(inInfoFile2 >> Credit.str >> Credit.freq >> Credit.density 
		>> Credit.Sequence >>Credit.FileCnt >>Credit.flag)
	{
		Copy(N[i],Credit);
		N[i].SetInfo = 1;
		N[i].flag  = Info[1].ClassFlag;
		i++;
	}
	
	Record = Unique(N,i);  //计算熵之前，先排除重复的记录并由Record记录无重复的数量

	WordsNumber[1] = (double)Record ;  //记录正常邮件得到的词汇总数	
	int Cnt = WordsNumber[0] + WordsNumber[1];    //由垃圾词库和正常词库得到的总的词汇数量
	double Density[2];  //Density[0]垃圾邮件的概率
	                    //Density[1]正常邮件的概率
	for (i = 0; i < 2 ;i ++)
	{
		Density[i] = (double)(Info[i].FileSum /FileNumber0_1);
	}

	double mGain = 0;

	int Max = (int)WordsNumber[0] + WordsNumber[1];
	WordGain *wordgainArray = new WordGain[Max];

	int m = 0,n = 0;
	for(int l = 0;l< Max; l++)
	{
		memset(wordgainArray[l].str,0,40);
		if(m < WordsNumber[0] && n < WordsNumber[1] && l < Max)
		{
			if(strcmp(M[m].str,N[n].str) < 0)
			{
				strcpy(wordgainArray[l].str,M[m].str);

				wordgainArray[l].density0 = M[m].density;
				wordgainArray[l].FileCnt0 = M[m].FileCnt;
				wordgainArray[l].freq0 = M[m].freq;
				wordgainArray[l].flag0 = 0;

				wordgainArray[l].density1 = 0.0;
				wordgainArray[l].FileCnt1 = 0;
				wordgainArray[l].freq1 = 0;
				wordgainArray[l].flag1 = 1;

				wordgainArray[l].gain = 0;	
				
				m ++;
			}
			else if(strcmp(M[m].str,N[n].str) == 0)
			{
				strcpy(wordgainArray[l].str,M[m].str);

				wordgainArray[l].density0 = M[m].density;
				wordgainArray[l].FileCnt0 = M[m].FileCnt;
				wordgainArray[l].freq0 = M[m].freq;
				wordgainArray[l].flag0 = 0;

				wordgainArray[l].density1 = N[n].density;
				wordgainArray[l].FileCnt1 = N[n].FileCnt;
				wordgainArray[l].freq1 = N[n].freq;
				wordgainArray[l].flag1 = 1;

				wordgainArray[l].gain = 0;	

				m ++;
				n ++;
			}
			else
			{
				strcpy(wordgainArray[l].str,M[m].str);

				wordgainArray[l].density0 = 0.0;
				wordgainArray[l].FileCnt0 = 0;
				wordgainArray[l].freq0 = 0;
				wordgainArray[l].flag0 = 0;

				wordgainArray[l].density1 = N[n].density;
				wordgainArray[l].FileCnt1 = N[n].FileCnt;
				wordgainArray[l].freq1 = N[n].freq;
				wordgainArray[l].flag1 = 1;

				wordgainArray[l].gain = 0;	

				n ++;
			}
		}
		else
		if( m < WordsNumber[0]  && n >= WordsNumber[1])
		{
			strcpy(wordgainArray[l].str,M[m].str);
			
			wordgainArray[l].density0 = M[m].density;
			wordgainArray[l].FileCnt0 = M[m].FileCnt;
			wordgainArray[l].freq0 = M[m].freq;
			wordgainArray[l].flag0 = 0;
			
			wordgainArray[l].density1 = 0.0;
			wordgainArray[l].FileCnt1 = 0;             
			wordgainArray[l].freq1 = 0;
			wordgainArray[l].flag1 = 1;
			
			wordgainArray[l].gain = 0;	
			m ++;
		}
		else 
		if( n < WordsNumber[1] && m >= WordsNumber[0])
		{
			strcpy(wordgainArray[l].str,N[n].str);
			
			wordgainArray[l].density0 = 0.0;
			wordgainArray[l].FileCnt0 = 0;
			wordgainArray[l].freq0 = 0;
			wordgainArray[l].flag0 = 0;
			
			wordgainArray[l].density1 = N[n].density;
			wordgainArray[l].FileCnt1 = N[n].FileCnt;
			wordgainArray[l].freq1 = N[n].freq;
			wordgainArray[l].flag1 = 1;
			
			wordgainArray[l].gain = 0;	

			n ++;
		}
		if(m >= WordsNumber[0] && n >= WordsNumber[1] )
			break;
	}

    int ReturnValue = l + 1;    //记录合并后的单词数目

	/************************************************************************/
     //计算增益
	double tempdensity[2] = {0.0,0.0};    //存储某个类别的单词在整个词库中的概率
	double TempWord = 0.0;
	for(i = 0; i < ReturnValue; i ++)
	{
		//记录正常邮件和垃圾邮件之和double FileNumber0_1	
			tempdensity[0] =(double) ((wordgainArray[i].FileCnt0 + 1.00) / (2.00 + FileNumber0_1)); 
			tempdensity[1] =(double) ((wordgainArray[i].FileCnt1 + 1.00) / (2.00 + FileNumber0_1)); 
			TempWord = (wordgainArray[i].FileCnt0 + wordgainArray[i].FileCnt1) * 1.00;


			mGain = tempdensity[0] * 
				log(tempdensity[0] * (1.00 / Density[0]) * ( 1.00 / ( TempWord / FileNumber0_1 ))) 
				/ log(2.000)
				+ 
				(1.00 - tempdensity[0]) *  
				log((1.00 - tempdensity[0]) * (1 / Density[0]) * ( 1.00 / (1.00 - TempWord / FileNumber0_1 ) )) 
				/ log(2.000)
				+
				tempdensity[1] * 
				log(tempdensity[1] * (1.00 / Density[1]) * ( 1.00 / ( TempWord / FileNumber0_1 ))) 
				/ log(2.000)
				+
				(1.00 - tempdensity[1]) *  
				log((1.00 - tempdensity[1]) * (1 / Density[1]) * ( 1.00 / (1.00 - TempWord / FileNumber0_1 ) )) 
				/ log(2.000);


			
			wordgainArray[i].gain = mGain;
	}

	//计算先验概率
	for(i = 0; i < ReturnValue; i ++)
	{
		strcpy(feature[i].str,wordgainArray[i].str);
		feature[i].UnNormalMail = 0;
		feature[i].UnNormalMail_Prior = (1.00 + wordgainArray[i].FileCnt0) / (2.00 + Info[0].FileSum);
		feature[i].NormalMail = 1;
		feature[i].NormalMail_Prior = (1.00 + wordgainArray[i].FileCnt1) / (2.00 + Info[1].FileSum);
		feature[i].gain = wordgainArray[i].gain;
	}
	

	if(Output == NULL)
		return -1;

	char tempstr[100];
	memset(tempstr,0,100);
	strcpy(tempstr,m_szInitDir);
	strcat(tempstr,Output);
	ofstream outFile(tempstr,ios::out| ios::app);
	if(!outFile)
	{
		cerr << "Output File could not be opened."  << endl;
		exit(1);
	}

	for( i = 0; i <= l; i++)
	{		
		outFile << feature[i].str << ' ' << feature[i].gain << ' ' 
			<< feature[i].NormalMail << ' ' << feature[i].NormalMail_Prior	<< ' ' 
			<< feature[i].UnNormalMail << ' ' << feature[i].UnNormalMail_Prior << '\n';
	}
	delete wordgainArray;
	delete N;
	delete M;

	return ReturnValue;
}

int CBrowseDir::ShellSortInFreq(Item M[], int N)    //按在文章中出现的次数排列
{	
	int gap = 0;
	int n = N;
	for(gap = n / 2; gap > 0; gap /= 2)
		for(int i = gap; i < n; i++)
			for(int j = i - gap; j >= 0; j -= gap)	
			{
				if(M[j].freq < M[j + gap].freq)
				{
					Item x;
					Copy(x,M[j]);
					Copy(M[j],M[j + gap]);
					Copy(M[j + gap],x);
				}
			}
			
			return 0;
}


int CBrowseDir::GetSignal()
{
	int m;
	cout <<"请输入m_switch的值(0 表示开始训练或 1表示启动反馈)：";
		cin >> m;
	if(m != 0 && m != 1){
			cout <<"输入m_switch的值(0 或 1)错误。";
		return -1;
	}
	m_switch = m;
	return 0;
}