word.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 "Word.h"

Word::Word(char *str)
{
    nameList = (nameNode *)malloc(sizeof(nameNode));
    nameList->name = new char[strlen(str)+1];
    nameList->next = NULL;
    strcpy(nameList->name,str);
    left = NULL;
    right = NULL;
    occurrence = 0;
    isFeat = 0;
    docList = NULL;
    docListTail = NULL; 
}

Word::~Word()
{
}

int Word::stemStrcmp(char *str)
{
    int r,s,t;
    nameNode *ptr;

    ptr = nameList;
    r = 1;
    while(ptr!=NULL && r!=0){
        r = strcmp(ptr->name,str);
        ptr = ptr->next;
    }

    ptr = nameList;
    s = 1;
    while(ptr!=NULL && r!=0 && s!=0){
        s = namecmp(ptr->name,str);
        ptr = ptr->next;
    }

    if(r == 0){
	t = 0;
    }else if(s == 0){
        nameNode *temp;
        temp = (nameNode *)malloc(sizeof(nameNode));
        temp->name = new char[strlen(str)+1];
        strcpy(temp->name,str);
        temp->next = nameList;
        nameList = temp;
	t = 0;
    }else if(s != 0){
	t = s;
    }

    return t;
}

int Word::namecmp(char *name,char *str)
{
  int result;
  result = strcmp(name,str);

  if(result != 0){

    int i,len1,len2,len,l;
    double v,avelen;

    len1 = strlen(name);
    len2 = strlen(str);
    if(len1 > len2){
        len = len2;
    }else{
        len = len1;
    }

    l = 0;
    i = 3;
    while(i<=len && l==0){
        if(strncmp(name,str,i) == 0){
            i++;
        }else{
            l = 1;
        }
    }

    i--;
    avelen = (len1+len2)/2.0;
    v = ((double) i)/avelen;
    if(i>=3 && v>Threshold){
        result = 0;
    }

  }

    return result;
}