📄 eclat.cc,v
字号:
head 1.3;access;symbols;locks zaki:1.3; strict;comment @// @;1.3date 2001.09.22.19.35.24; author zaki; state Exp;branches;next 1.2;1.2date 2001.06.24.21.17.20; author zaki; state Exp;branches;next 1.1;1.1date 2001.06.12.16.41.58; author zaki; state Exp;branches;next ;desc@Charm with Hashing.@1.3log@*** empty log message ***@text@#include <iostream.h>#include <stdlib.h>#include "eclat.h"#include "assoc.h"#include "Graph.h"#include "Itemset.h"#include "memman.h"#include "Array.h"//#include "gbk.h"#include "Util.h"//#include "cliques.h"#include "maxset.h"#include "defset.h"#include "Count.h"#include "hashtable.h"#include "stats.h"extern unsigned long sumtidlist;extern int DBASE_MAXITEM;boolean ECLAT_bfs=TRUE;boolean ECLAT_eqc=TRUE;int ECLAT_search=BOTUP;#define ECLAT_maxisetsz 100CountAry FreCount, CloCount;Itemset *Eqclass::iset = NULL;Array<int> NumLargeItemset;int maxiter = 2;maxset *MaxSet;HashTable *HT;cHashTable *cHT;void Eqclass::alloc_tmpiset(int sz){ iset = new Itemset(ECLAT_maxisetsz,sz);}#ifdef PRUNEboolean prune_test(Eqclass *EQ) { ListNodes<Itemset *> *tempnode=EQ->head(); Array<int> *tempAry=new Array<int>(tempnode->item()->iset()); for(tempnode=tempnode->next();tempnode;tempnode=tempnode->next()) { tempAry->add(tempnode->item()->litem()); } boolean TF=MaxSet->subset(tempAry); delete tempAry; return TF;}#endif//calculate totmem for elements of an eqclass,//if add is true we add to TOTMEM, else we substractvoid getmem(Eqclass *cluster, boolean add){ long sum=0; ListNodes<Itemset *> *hdr; for (hdr = cluster->head(); hdr; hdr=hdr->next()){ sum += hdr->item()->tidsize(); } if (add) TOTMEM += sum; else TOTMEM -= sum; cout << TOTMEM << endl; //print out the tot mem}//herevoid process_cluster(Eqclass *cluster, int iter){ if (memflg) getmem(cluster, TRUE); if (maxiter < iter) maxiter = iter; //cout << "cluster:" << *cluster ; ListNodes<Itemset *> *hdr, *hdr2,*phdr2,*tempnode; int subset; tempnode=cluster->head(); for (hdr = cluster->head(); hdr; hdr=hdr->next()){ Eqclass *EQ = new Eqclass; phdr2 =hdr; sumtidlist += hdr->item()->tidsize(); for (hdr2=hdr->next(); hdr2 ; ){ sumtidlist += hdr2->item()->tidsize(); Itemset *temp; //cout << "hdr->item" << *hdr->item() ; //cout << "hdr->item()->tid():" << *hdr->item()->tidlist() << endl; //cout << "hdr2->item()" << *hdr2->item() ; //cout << "hdr2->item()->tid():" << *hdr2->item()->tidlist() << endl; if (use_diff){ if (!diff_input && !use_diff_f2 && iter == 2) temp=Itemset::diff2it(hdr->item(), hdr2->item(),&subset); else temp=Itemset::diff(hdr->item(), hdr2->item(),0,&subset); } else temp=Itemset::intersect(hdr->item(), hdr2->item(),0,&subset); if (hdr->item()->isetsize() > 1) stats.incrcand(hdr->item()->isetsize()); if (temp){ FreCount.add(temp->iset()->size()); stats.incravgtid(temp->iset()->size()-1,temp->tidsize()); switch(subset){ case AeqB: //A=B, A=AB, & delete B //cout << "subset: " << subset << " A=B" << endl; //cout << "hdr->item" << *hdr->item() ; //cout << "hdr->item()->tid():" << *hdr->item()->tidlist() << endl; //cout << "hdr2->item()" << *hdr2->item() ; //cout << "hdr2->item()->tid():" << *hdr2->item()->tidlist() << endl; if (use_diff){ //hdr->item()->insert(hdr2->item()->litem()); hdr->item()->copy(temp->iset(), NULL); //hdr->item()->hval() = temp->hval(); //temp->copy(NULL, hdr->item()->tidlist()); //temp->support() = hdr->item()->support(); //temp->diff() = hdr->item()->diff(); //temp->idrop() = hdr->item()->idrop(); } else{ //delete hdr->item(); //hdr->item()=new Itemset(*temp); //hdr->item()->support() = temp->support(); //hdr->item()->diff() = temp->diff(); hdr->item()->insert(hdr2->item()->litem()); delete temp; } for(tempnode= EQ->head(); tempnode; tempnode=tempnode->next()){ //replace A with AB tempnode->item()->insert(hdr2->item()->litem()); FreCount.add(tempnode->item()->iset()->size()); stats.incravgtid(tempnode->item()->iset()->size()-1, tempnode->item()->tidsize()); } if(phdr2){ phdr2->lnext()=hdr2->next(); delete hdr2; hdr2=phdr2; } else { cluster->set_head(hdr2->next()); delete hdr2; hdr2=NULL; } //cout << "after A=B" << *hdr->item() << endl; break; case AsubB: // A<B,replace A with AB //cout << "subset: " << subset << " A<B" << endl; //cout << "temp" << *temp ; //hdr->set_bit(); if (use_diff){ //hdr->item()->insert(hdr2->item()->litem()); hdr->item()->copy(temp->iset(), NULL); //hdr->item()->hval() = temp->hval(); //temp->copy(NULL, hdr->item()->tidlist()); //temp->support() = hdr->item()->support(); //temp->diff() = hdr->item()->diff(); //temp->idrop() = hdr->item()->idrop(); } else { //delete hdr->item(); //hdr->item()=new Itemset(*temp); //hdr->item()->support() = temp->support(); //hdr->item()->diff() = temp->diff(); hdr->item()->insert(hdr2->item()->litem()); delete temp; } tempnode=EQ->head(); for(; tempnode; tempnode=tempnode->next()){ //replace A with AB tempnode->item()->insert(hdr2->item()->litem()); FreCount.add(tempnode->item()->iset()->size()); stats.incravgtid(tempnode->item()->iset()->size()-1, tempnode->item()->tidsize()); } //cout << "hdr: " << *hdr->item() << endl; break; case BsubA: // A>B, delete B & save AB //cout << "subset: " << subset << " A>B" << endl; //hdr->set_bit(); if(phdr2) { phdr2->lnext()=hdr2->next(); delete hdr2; hdr2=phdr2; } else { cluster->set_head(hdr2->next()); delete hdr2; hdr2=NULL; } break; case AneqB: //A<>B,nothing //cout << "subset: " << subset << " A<>B" << endl; //hdr->set_bit(); //hdr2->set_bit(); break; } // edn of switch //cout << "join: " << *temp ; if(subset == BsubA || subset == AneqB ){ if (process_order == DE_SUP) EQ->addsort(temp,Itemset::cmp_sup,0); else if (process_order == IN_SUP) EQ->addsort(temp,Itemset::cmp_sup,1); else EQ->append(temp); } //end of if (subset>=2) } //end if temp if(hdr2) { phdr2=hdr2; hdr2=hdr2->next(); } else { cout << "ERROR" << endl; //hdr2=cluster->head(); } }// end of hdr2 bool flg = true; if(use_hash){ if(!HT->add(hdr->item())) flg = false; } else if (use_hash_map){ if (!cHT->add(hdr->item())) flg = false; } if (flg){ CloCount.add(hdr->item()->iset()->size()); if (use_output) cout << *hdr->item(); } if (EQ->size() > 1){ process_cluster(EQ, iter+1); delete EQ; } else { if(EQ->size()==1) { flg = true; if(use_hash){ if (!HT->add(EQ->head()->item())) flg = false; } else if (use_hash_map){ if (!cHT->add(EQ->head()->item())) flg = false; } if (flg){ CloCount.add(EQ->head()->item()->iset()->size()); if (use_output) cout << *EQ->head()->item(); } } if(EQ) delete EQ; } //end of EQ->size() } // end of hdr/* for(tempnode=cluster->head();tempnode ;tempnode=tempnode->next()) { if(!tempnode->chbit()){ CloCount.add(tempnode->item()->iset()->size()); cout << *tempnode->item() ; } }*/ if (memflg) getmem(cluster, FALSE);}void process_F1(Eqclass *F1EQ) { int iter =1; ListNodes<Itemset *> *hdr, *hdr2,*phdr2,*tempnode; int subset; Array <int> *cls=new Array<int>; Itemset *it1; hdr = F1EQ->head(); for (; hdr; hdr=hdr->next()){
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -