dc.cc

一个增量文本聚类的算法。 参考文献： Wai-chiu Wong, Ada Wai-chee Fu, Incremental Document Clustering for Web Page Cl

#include <stdio.h>
#include <stdlib.h>
#include <ctype.h>
#include <string.h>
#include <iostream.h>
#include "DC.h"

DC::DC(int id,int *fv,int nf,char *lbl)
{
    CID = id;
    if(id >= 0){
    	N = 1;
    	numFeat = nf;
    	W = new int[nf];
    	for(int i=0;i<nf;i++){
	    W[i] = fv[i];
    	}
    	docList = (dNode *)malloc(sizeof(dNode));
    	docList->ID = id;
    	strcpy(docList->label,lbl); 
    	docList->next = NULL;
    }else{
    	N = 0;
    	numFeat = nf;
    	W = new int[nf];
    	for(int i=0;i<nf;i++){
	    W[i] = 0;
    	}
    	docList = NULL;
    }
}

DC::~DC()
{
}

void DC::merge(DC *ent)
{
    if(N > 0){
    	N = N + ent->N;
    	for(int i=0;i<numFeat;i++){
	    W[i] = W[i] + ent->W[i];
    	}
    	dNode *dlc,*dlp;
    	dlc = docList;
    	dlp = docList;
    	while(dlc != NULL){
	    dlp = dlc;
	    dlc = dlc->next;
    	}
	dNode *dla,*dlb;
	dla = dlp;
	dlb = ent->docList;
	while(dlb != NULL){
            dla->next = (dNode *)malloc(sizeof(dNode));
	    dla->next->ID = dlb->ID;
	    strcpy(dla->next->label,dlb->label);
	    dla->next->next = NULL; 
	    dla = dla->next;
	    dlb = dlb->next;
	}
    }else{
    	N = ent->N;
    	for(int i=0;i<numFeat;i++){
	    W[i] = ent->W[i];
        }
	dNode *dla,*dlb;
	dlb = ent->docList;
        docList = (dNode *)malloc(sizeof(dNode));
	docList->ID = dlb->ID;
	strcpy(docList->label,dlb->label);
	docList->next = NULL; 
	dla = docList;
	dlb = dlb->next;
	while(dlb != NULL){
            dla->next = (dNode *)malloc(sizeof(dNode));
	    dla->next->ID = dlb->ID;
	    strcpy(dla->next->label,dlb->label);
	    dla->next->next = NULL; 
	    dla = dla->next;
	    dlb = dlb->next;
	}
	//docList->ID = ent->docList;
    }
}

void DC::showInfo()
{
    dNode *dl;
    dl = docList;
    while(dl != NULL){
        printf("%d:%s ",dl->ID,dl->label);
	dl = dl->next;
    }
    printf("\n");
    for(int i=0;i<numFeat;i++){
	printf("%d ",W[i]);
    }
    printf("\n");
}

int DC::showStat(char *lbl)
{
    int n;
    
    dNode *dl;
    dl = docList;
    n = 0;
    while(dl != NULL){
        if(strcmp(lbl,dl->label) == 0){
	    n++;
	}
        dl = dl->next;
    }
    
    return n;
}