C++ TIntH::IsKey方法代码示例

// burn each link independently (forward with FwdBurnProb, backward with BckBurnProb)
void TForestFire::BurnExpFire() {
  const double OldFwdBurnProb = FwdBurnProb;
  const double OldBckBurnProb = BckBurnProb;
  const int NInfect = InfectNIdV.Len();
  const TNGraph& G = *Graph;
  TIntH BurnedNIdH;               // burned nodes
  TIntV BurningNIdV = InfectNIdV; // currently burning nodes
  TIntV NewBurnedNIdV;            // nodes newly burned in current step
  bool HasAliveNbrs;              // has unburned neighbors
  int NBurned = NInfect, NDiedFire=0;
  for (int i = 0; i < InfectNIdV.Len(); i++) {
    BurnedNIdH.AddDat(InfectNIdV[i]); }
  NBurnedTmV.Clr(false);  NBurningTmV.Clr(false);  NewBurnedTmV.Clr(false);
  for (int time = 0; ; time++) {
    NewBurnedNIdV.Clr(false);
    // for each burning node
    for (int node = 0; node < BurningNIdV.Len(); node++) {
      const int& BurningNId = BurningNIdV[node];
      const TNGraph::TNodeI Node = G.GetNI(BurningNId);
      HasAliveNbrs = false;
      NDiedFire = 0;
      // burn forward links  (out-links)
      for (int e = 0; e < Node.GetOutDeg(); e++) {
        const int OutNId = Node.GetOutNId(e);
        if (! BurnedNIdH.IsKey(OutNId)) { // not yet burned
          HasAliveNbrs = true;
          if (Rnd.GetUniDev() < FwdBurnProb) {
            BurnedNIdH.AddDat(OutNId);  NewBurnedNIdV.Add(OutNId);  NBurned++; }
        }
      }
      // burn backward links (in-links)
      if (BckBurnProb > 0.0) {
        for (int e = 0; e < Node.GetInDeg(); e++) {
          const int InNId = Node.GetInNId(e);
          if (! BurnedNIdH.IsKey(InNId)) { // not yet burned
            HasAliveNbrs = true;
            if (Rnd.GetUniDev() < BckBurnProb) {
              BurnedNIdH.AddDat(InNId);  NewBurnedNIdV.Add(InNId);  NBurned++; }
          }
        }
      }
      if (! HasAliveNbrs) { NDiedFire++; }
    }
    NBurnedTmV.Add(NBurned);
    NBurningTmV.Add(BurningNIdV.Len() - NDiedFire);
    NewBurnedTmV.Add(NewBurnedNIdV.Len());
    //BurningNIdV.AddV(NewBurnedNIdV);   // node is burning eternally
    BurningNIdV.Swap(NewBurnedNIdV);    // node is burning just 1 time step
    if (BurningNIdV.Empty()) break;
    FwdBurnProb = FwdBurnProb * ProbDecay;
    BckBurnProb = BckBurnProb * ProbDecay;
  }
  BurnedNIdV.Gen(BurnedNIdH.Len(), 0);
  for (int i = 0; i < BurnedNIdH.Len(); i++) {
    BurnedNIdV.Add(BurnedNIdH.GetKey(i)); }
  FwdBurnProb = OldFwdBurnProb;
  BckBurnProb = OldBckBurnProb;
}

int Intersect(TUNGraph::TNodeI Node, TIntH NNodes){
  int br = 0;
  for (int i = 0; i<Node.GetDeg(); i++)
  {
    if (NNodes.IsKey(Node.GetNbrNId(i)))
      br++;
  }
  if (NNodes.IsKey(Node.GetId()))
    br++;

  return br;
}

// set Language and Country for movies that do not have the value set
// for every movie find the mojority language/country in 1-hop neighborhood and set it
void TImdbNet::SetLangCntryByMajority() {
  // set language
  while (true) {
    TIntPrV NIdToVal;
    for (TNodeI NI = BegNI(); NI < EndNI(); NI++) {
      if (NI().GetLang() != 0) { continue; }
      int Nbhs=0; TIntH LangCntH;
      for (int e = 0; e < NI.GetOutDeg(); e++) {
        LangCntH.AddDat(NI.GetOutNDat(e).GetLang()) += 1;  Nbhs++; }
      for (int e = 0; e < NI.GetInDeg(); e++) {
        LangCntH.AddDat(NI.GetInNDat(e).GetLang()) += 1;  Nbhs++; }
      if (LangCntH.IsKey(0)) { Nbhs-=LangCntH.GetDat(0); LangCntH.GetDat(0)=0; }
      LangCntH.SortByDat(false);
      if (LangCntH.GetKey(0) == 0) { continue; }
      if (LangCntH[0]*2 >= Nbhs) { 
        NIdToVal.Add(TIntPr(NI.GetId(), LangCntH.GetKey(0))); }
    }
    if (NIdToVal.Empty()) { break; } // done
    for (int i = 0; i < NIdToVal.Len(); i++) {
      GetNDat(NIdToVal[i].Val1).Lang = NIdToVal[i].Val2; }
    printf("  language set: %d\n", NIdToVal.Len());
  }
  int cnt=0;
  for (TNodeI NI = BegNI(); NI < EndNI(); NI++) { if (NI().GetLang()==0) cnt++; }
  printf(" NO language: %d\n\n", cnt);
  // set country
  while (true) {
    TIntPrV NIdToVal;
    for (TNodeI NI = BegNI(); NI < EndNI(); NI++) {
      if (NI().GetCntry() != 0) { continue; }
      int Nbhs=0; TIntH CntryCntH;
      for (int e = 0; e < NI.GetOutDeg(); e++) {
        CntryCntH.AddDat(NI.GetOutNDat(e).GetCntry()) += 1;  Nbhs++; }
      for (int e = 0; e < NI.GetInDeg(); e++) {
        CntryCntH.AddDat(NI.GetInNDat(e).GetCntry()) += 1;  Nbhs++; }
      if (CntryCntH.IsKey(0)) { Nbhs-=CntryCntH.GetDat(0); CntryCntH.GetDat(0)=0; }
      CntryCntH.SortByDat(false);
      if (CntryCntH.GetKey(0) == 0) { continue; }
      if (CntryCntH[0]*2 >= Nbhs) { 
        NIdToVal.Add(TIntPr(NI.GetId(), CntryCntH.GetKey(0))); }
    }
    if (NIdToVal.Empty()) { break; } // done
    for (int i = 0; i < NIdToVal.Len(); i++) {
      GetNDat(NIdToVal[i].Val1).Cntry = NIdToVal[i].Val2; }
    printf("  country set: %d\n", NIdToVal.Len());
  }
  cnt=0;
  for (TNodeI NI = BegNI(); NI < EndNI(); NI++) { if (NI().GetCntry()==0) cnt++; }
  printf(" NO country: %d\n\n", cnt);
}

//////////////////////////////////////////////////////////////////////////
// String-To-Words
void TStrParser::DocStrToWIdV(const TStr& _DocStr, TIntV& WordIdV, const bool& Stemm) {
    TStr DocStr = _DocStr.GetUc();  // to upper case
    TStrV WordV; DocStr.SplitOnWs(WordV); int WordN = WordV.Len();
    WordIdV.Reserve(WordN, 0);

    PStemmer Stemmer = TStemmer::New(stmtPorter);
    TIntH WordsInDoc;
    for (int WordC = 0; WordC < WordN; WordC++) {
        TStr WordStr;
        if (Stemm) {
            WordStr = Stemmer->GetStem(WordV[WordC]);
        } else {
            WordStr = WordV[WordC];
        }
        int WId = GetWId(WordStr);
        if (WId == -1) {
            WId = WordToIdH.AddKey(WordStr);
            WordToIdH[WId] = 0;
        }
        WordIdV.Add(WId);
        
        // is it first time we see this word in this doc?
        if (!WordsInDoc.IsKey(WId)) WordsInDoc.AddKey(WId);
    }

    //do some statistics for DF
    DocsParsed++;
    for (int i = 0, l = WordsInDoc.Len(); i < l; i++)
        WordToIdH[WordsInDoc.GetKey(i)]++;

    Assert(WordV.Len() == WordIdV.Len());
}

// set actor's language and country
void TImdbNet::SetActorCntryLangByMajority() {
    // set language
  TIntPrV NIdToVal;
  for (TNodeI NI = BegNI(); NI < EndNI(); NI++) {
    if (! NI().IsActor()) { continue; }
    IAssert(NI().GetLang() == 0); // no language set
    IAssert(NI.GetInDeg() == 0);  // actors point to movies
    int Nbhs=0;  TIntH LangCntH;
    for (int e = 0; e < NI.GetOutDeg(); e++) {
      LangCntH.AddDat(NI.GetOutNDat(e).GetLang()) += 1;  Nbhs++; }
    if (LangCntH.IsKey(0)) { Nbhs-=LangCntH.GetDat(0); LangCntH.GetDat(0)=0; }
    LangCntH.SortByDat(false);
    if (LangCntH.GetKey(0) == 0) { continue; }
    if (LangCntH[0]*2 >= Nbhs) { 
      NIdToVal.Add(TIntPr(NI.GetId(), LangCntH.GetKey(0))); }
  }
  for (int i = 0; i < NIdToVal.Len(); i++) {
    GetNDat(NIdToVal[i].Val1).Lang = NIdToVal[i].Val2; }
  printf("  language set: %d\n", NIdToVal.Len());
  
  int cnt=0;
  for (TNodeI NI = BegNI(); NI < EndNI(); NI++) { if (NI().IsActor() && NI().GetLang()==0) cnt++; }
  printf("  Actors NO language: %d\n\n", cnt);
  // set country
  NIdToVal.Clr(true);
  for (TNodeI NI = BegNI(); NI < EndNI(); NI++) {
    if (! NI().IsActor()) { continue; }
    IAssert(NI().GetCntry() == 0); // no country set
    IAssert(NI.GetInDeg() == 0);   // actors point to movies
    int Nbhs=0; TIntH CntryCntH;
    for (int e = 0; e < NI.GetOutDeg(); e++) {
      CntryCntH.AddDat(NI.GetOutNDat(e).GetCntry()) += 1;  Nbhs++; }
    if (CntryCntH.IsKey(0)) { Nbhs-=CntryCntH.GetDat(0); CntryCntH.GetDat(0)=0; }
    CntryCntH.SortByDat(false);
    if (CntryCntH.GetKey(0) == 0) { continue; }
    if (CntryCntH[0]*2 >= Nbhs) { 
      NIdToVal.Add(TIntPr(NI.GetId(), CntryCntH.GetKey(0))); }
  }
  for (int i = 0; i < NIdToVal.Len(); i++) {
    GetNDat(NIdToVal[i].Val1).Cntry = NIdToVal[i].Val2; }
  printf("  country set: %d\n", NIdToVal.Len());
  cnt=0;
  for (TNodeI NI = BegNI(); NI < EndNI(); NI++) { if (NI().IsActor() && NI().GetCntry()==0) cnt++; }
  printf("  Actors NO country: %d\n\n", cnt);
}

void TFtrGenSparseNumeric::Add(const TStr& Str, TIntFltKdV& SpV, int& Offset) const {
    TStrV EltV; Str.SplitOnAllCh(';', EltV); TIntH UsedIdH;
    for (int EltN = 0; EltN < EltV.Len(); EltN++) {
        int Id; TStr Val; Split(EltV[EltN], Id, Val);
        EAssertR(!UsedIdH.IsKey(Id), "Field ID repeated in '" + Str + "'!");
        int TmpOffset = Offset + Id;
        FtrGen->Add(Val, SpV, TmpOffset);
        UsedIdH.AddKey(Id);
    }
    Offset += GetVals();
}

double GetGroupDegreeCentr0(const PUNGraph& Graph, const TIntH& GroupNodes) {
  int deg;
  TIntH NN;
  for (int i = 0; i<GroupNodes.Len(); i++) {
    deg = Graph->GetNI(GroupNodes.GetDat(i)).GetDeg();
    for (int j = 0; j < deg; j++) {
      if (GroupNodes.IsKey(Graph->GetNI(GroupNodes.GetDat(i)).GetNbrNId(j)) == 0)
        NN.AddDat(Graph->GetNI(GroupNodes.GetDat(i)).GetNbrNId(j), GroupNodes.GetDat(i));
    }
  }
  return (double)NN.Len();
}

// network cascade: add spurious edges
// for more details see "Correcting for Missing Data in Information Cascades" by E. Sadikov, M. Medina, J. Leskovec, H. Garcia-Molina. WSDM, 2011
PNGraph AddSpuriousEdges(const PUNGraph& Graph, const PNGraph& Casc, TIntH NIdTmH) {
  TIntPrV EdgeV;
  for (TNGraph::TNodeI NI = Casc->BegNI(); NI < Casc->EndNI(); NI++) {
    TUNGraph::TNodeI GNI = Graph->GetNI(NI.GetId());
    const int Tm = NIdTmH.GetDat(NI.GetId());
    for (int i=0,j=0; i < GNI.GetOutDeg(); i++) {
      const int Dst = GNI.GetOutNId(i);
      if (NIdTmH.IsKey(Dst) && Tm<NIdTmH.GetDat(Dst) && ! NI.IsNbhNId(Dst)) {
        EdgeV.Add(TIntPr(GNI.GetId(), Dst)); }
    }
  }
  PNGraph NetCasc = TNGraph::New();
  *NetCasc = *Casc;
  for (int e = 0; e < EdgeV.Len(); e++) {
    NetCasc->AddEdge(EdgeV[e].Val1, EdgeV[e].Val2); }
  return NetCasc;
}

double GetGroupDegreeCentr(const PUNGraph& Graph, const TIntH& GroupNodes) {
  int deg;
  TIntH NN;
  TIntH GroupNodes1;

  for (THashKeyDatI<TInt, TInt> NI = GroupNodes.BegI(); NI < GroupNodes.EndI(); NI++)
    GroupNodes1.AddDat(NI.GetDat(), NI.GetDat());

  for (THashKeyDatI<TInt, TInt> NI = GroupNodes1.BegI(); NI < GroupNodes1.EndI(); NI++){
    TUNGraph::TNodeI node = Graph->GetNI(NI.GetKey());
    deg = node.GetDeg();
    for (int j = 0; j < deg; j++){
      if (GroupNodes1.IsKey(node.GetNbrNId(j)) == 0 && NN.IsKey(node.GetNbrNId(j)) == 0)
        NN.AddDat(node.GetNbrNId(j), NI.GetKey());
    }
  }

  return (double)NN.Len();
}

// simulate SI model cascade using infection probability Beta until the cascade stops or reaches size MxCascSz
PNGraph RunSICascade2(PUNGraph G, const double& Beta, const int& MxCascSz, TIntH& NIdInfTmH) {
  PNGraph Casc = TNGraph::New();
  const int StartNId = G->GetRndNId();
  Casc->AddNode(StartNId);
  NIdInfTmH.AddDat(StartNId, NIdInfTmH.Len());
  TIntQ Q; Q.Push(StartNId);
  while (! Q.Empty()) {
    const TUNGraph::TNodeI NI = G->GetNI(Q.Top()); Q.Pop();
    for (int i = 0; i < NI.GetOutDeg(); i++) {
      if (TInt::Rnd.GetUniDev() < Beta && ! NIdInfTmH.IsKey(NI.GetOutNId(i))) {
        Casc->AddNode(NI.GetOutNId(i));
        NIdInfTmH.AddDat(NI.GetOutNId(i), NIdInfTmH.Len());
        Casc->AddEdge(NI.GetId(), NI.GetOutNId(i));
        if (Casc->GetNodes() == MxCascSz) { return Casc; }
        Q.Push(NI.GetOutNId(i));
      }
    }
  }
  return Casc;
}

PLwOntoGround TLwOntoGround::GetOntoGround(
 const PLwOnto& LwOnto, const PBowDocBs& BowDocBs,
 const TStr& LangNm, const bool& DocCatIsTermIdP,
 const double& CutWordWgtSumPrc){
  printf("Generating Ontology-Classifier...\n");
  // shortcuts
  PLwTermBs TermBs=LwOnto->GetTermBs();
  int Terms=TermBs->GetTerms();
  PLwLinkBs LinkBs=LwOnto->GetLinkBs();
  PLwLinkTypeBs LinkTypeBs=LwOnto->GetLinkTypeBs();
  int LangId=LwOnto->GetLangBs()->GetLangId(LangNm);
  int Docs=BowDocBs->GetDocs();
  // create tfidf
  printf("  Creating BowDocWgtBs ...");
  PBowDocWgtBs BowDocWgtBs=TBowDocWgtBs::New(BowDocBs, bwwtNrmTFIDF);
  PBowSim BowSim=TBowSim::New(bstCos);
  printf(" Done.\n");
  // collect documents per ontology-term
  printf("  Collecting documents per ontology-term ...\n");
  TIntIntVH TermIdToDIdVH; int PosCats=0; int NegCats=0;
  for (int DId=0; DId<Docs; DId++){
    printf("    Docs:%d/%d Pos:%d Neg:%d\r", 1+DId, Docs, PosCats, NegCats);
    for (int DocCIdN=0; DocCIdN<BowDocBs->GetDocCIds(DId); DocCIdN++){
      // get document-category
      int CId=BowDocBs->GetDocCId(DId, DocCIdN);
      TStr CatNm=BowDocBs->GetCatNm(CId);
      // get term-id
      if (DocCatIsTermIdP){
        int TermId=CatNm.GetInt();
        if (TermBs->IsTermId(TermId)){
          TermIdToDIdVH.AddDat(TermId).Add(DId); PosCats++;
        } else {NegCats++;}
      } else {
        if (TermBs->IsTermId(CatNm, LangId)){
          int TermId=TermBs->GetTermId(CatNm, LangId);
          TermIdToDIdVH.AddDat(TermId).Add(DId); PosCats++;
        } else {NegCats++;}
      }
    }
  }
  printf("    Docs:%d/%d Pos:%d Neg:%d\n", Docs, Docs, PosCats, NegCats);
  printf("  Done.\n");
  // create sub-terms & up-terms vectors
  printf("  Creating sub-terms & up-terms vectors ...");
  TIntIntVH Const_TermIdToSubTermIdVH;
  TIntIntVH TermIdToSubTermIdVH;
  TIntIntVH TermIdToUpTermIdVH;
  for (int TermN=0; TermN<Terms; TermN++){
    int TermId=TermBs->GetTermId(TermN);
    for (int LinkN=0; LinkN<LinkBs->GetFromLinks(TermId); LinkN++){
      int LinkTypeId; int DstTermId;
      LinkBs->GetFromLink(TermId, LinkN, LinkTypeId, DstTermId);
      TStr LinkTypeNm=LinkTypeBs->GetLinkType(LinkTypeId)->GetLinkTypeNm();
      if (LinkTypeNm=="NT"){
        Const_TermIdToSubTermIdVH.AddDat(TermId).Add(DstTermId);
        TermIdToSubTermIdVH.AddDat(TermId).Add(DstTermId);
        TermIdToUpTermIdVH.AddDat(DstTermId).Add(TermId);
      }
    }
  }
  printf("   Done.\n");
  // create centroids
  printf("  Creating centroids ...\n");
  THash<TInt, PBowSpV> TermIdToConceptSpVH;
  TIntIntVH TermIdToSubTermDIdVH;
  TIntH ProcTermIdH;
  int PrevActiveTerms=-1;
  forever{
    // count active nodes for processing
    int ActiveTerms=0;
    for (int TermN=0; TermN<Terms; TermN++){
      int TermId=TermBs->GetTermId(TermN);
      if ((TermIdToSubTermIdVH.IsKey(TermId))&&
       (TermIdToSubTermIdVH.GetDat(TermId).Len()>0)){
        ActiveTerms++;
      }
    }
    // stop if no change from previous round
    printf("    Active-Terms:%d\n", ActiveTerms);
    if (ActiveTerms==PrevActiveTerms){break;}
    PrevActiveTerms=ActiveTerms;
    // reduce active-nodes with zero-ancestors
    for (int TermN=0; TermN<Terms; TermN++){
      int TermId=TermBs->GetTermId(TermN);
      if (ProcTermIdH.IsKey(TermId)){continue;}
      if ((!TermIdToSubTermIdVH.IsKey(TermId))||
       (TermIdToSubTermIdVH.GetDat(TermId).Len()==0)){
        printf("    %d/%d\r", 1+TermN, Terms);
        ProcTermIdH.AddKey(TermId);
        // collect document-ids
        TIntV TermDIdV;
        if (TermIdToDIdVH.IsKey(TermId)){
          TermDIdV.AddV(TermIdToDIdVH.GetDat(TermId));}
        if (TermIdToSubTermDIdVH.IsKey(TermId)){
          TermDIdV.AddV(TermIdToSubTermDIdVH.GetDat(TermId));}
        // create concept-vector if any documents
        if (TermDIdV.Len()>0){
          PBowSpV ConceptSpV=
           TBowClust::GetConceptSpV(BowDocWgtBs, BowSim, TermDIdV, CutWordWgtSumPrc);
          TermIdToConceptSpVH.AddDat(TermId, ConceptSpV);
//.........这里部分代码省略.........

// load from allactors.zip that was prepared by Brad Malin in 2005
PImdbNet TImdbNet::LoadTxt(const TStr& ActorFNm) {
  PImdbNet Net = TImdbNet::New();
  TStrV ColV;
  char line [2024];
  int NLines=0, DupEdge=0, Year, Position, ActorNId, MovieNId;
  TIntH ActorNIdH;
  THash<TIntPr, TInt> MovieNIdH;
  FILE *F = fopen(ActorFNm.CStr(), "rt");  fgets(line, 2024, F);
  while (! feof(F)) {
    memset(line, 0, 2024);
    fgets(line, 2024, F);
    if (strlen(line) == 0) break;
    TStr(line).SplitOnAllCh('|', ColV, false);  IAssert(ColV.Len() == 7);
    const int NameStrId = Net->AddStr(ColV[0].GetTrunc().GetLc()+" "+ColV[1].GetTrunc().GetLc());
    const int MovieStrId = Net->AddStr(ColV[2].GetTrunc().GetLc());
    TStr YearStr = ColV[3].GetTrunc();
    if (YearStr.Len() > 4) YearStr = YearStr.GetSubStr(0, 3);
    Year = 1;  YearStr.IsInt(Year);
    const TMovieTy MovieTy = TImdbNet::GetMovieTy(ColV[4]);
    Position = TInt::Mx;  ColV[5].GetTrunc().IsInt(Position);
    IAssert(ColV[6].GetTrunc()[0] == 'M' || ColV[6].GetTrunc()[0]=='F');
    const bool IsMale = ColV[6].GetTrunc()[0] == 'M';
    // create nodes  
    if (ActorNIdH.IsKey(NameStrId)) { 
      ActorNId = ActorNIdH.GetDat(NameStrId); }
    else { 
      ActorNId = Net->AddNode(-1, TImdbNode(NameStrId, Year, Position, IsMale));
      ActorNIdH.AddDat(NameStrId, ActorNId);
    }
    if (MovieNIdH.IsKey(TIntPr(MovieStrId, Year))) {
      MovieNId = MovieNIdH.GetDat(TIntPr(MovieStrId, Year)); }
    else {
      MovieNId = Net->AddNode(-1, TImdbNode(NameStrId, Year, MovieTy)); 
      MovieNIdH.AddDat(TIntPr(MovieStrId, Year), MovieNId); 
    }
    if (! Net->IsEdge(ActorNId, MovieNId)) { 
      Net->AddEdge(ActorNId, MovieNId); }
    else { DupEdge++; }
    if (++NLines % 100000 == 0) printf("\r  %dk  ", NLines/1000);
  }
  fclose(F);
  printf("duplicate edges: %d\n", DupEdge);
  printf("nodes:  %d\n", Net->GetNodes());
  printf("edges:  %d\n", Net->GetEdges());
  printf("actors: %d\n", ActorNIdH.Len());
  printf("movies: %d\n", MovieNIdH.Len());
  // set the actor year to the year of his first movie
  int NUpdates=0;
  for (TNet::TNodeI NI = Net->BegNI(); NI < Net->EndNI(); NI++) {
    if (NI().IsActor()) {
      int MinYear = NI().GetYear();
      for (int e = 0; e < NI.GetOutDeg(); e++) {
        const TImdbNode& NodeDat = Net->GetNDat(NI.GetOutNId(e));
        if (NodeDat.IsMovie()) MinYear = TMath::Mn(MinYear, NodeDat.GetYear());
      }
      if (NI().Year != MinYear) NUpdates++;
      NI().Year = MinYear;
    }
  }
  printf("updated actor times: %d\n", NUpdates);
  return Net;
}

/////////////////////////////////////////////////
// NIST-score
double TEvalScoreNist::Eval(const PTransCorpus& TransCorpus, const TIntV& _SentIdV) {
    // check if the corpus has translations
    IAssert(TransCorpus->IsTrans());

    // ngram counts (cliped and full)
    TIntH ClipCountNGramH, CountNGramH;
    // ngram info score
    TIntFltH NGramInfoH;
    // candidate and effective reference length
    double FullTransLen = 0.0, FullRefLen = 0.0;

    // iterate over sentences
    TIntV SentIdV = _SentIdV;
    if (SentIdV.Empty()) { TransCorpus->GetSentIdV(SentIdV); }
    const int Sents = SentIdV.Len();
    for (int SentIdN = 0; SentIdN < Sents; SentIdN++) {
        const int SentId = SentIdV[SentIdN];
        // tokenize translation
        TIntV TransWIdV; Parse(TransCorpus->GetTransStr(SentId), TransWIdV);
        TIntH TransNGramH; GetNGramH(TransWIdV, MxNGramLen, TransNGramH);
        TIntH FreeTransNGramH = TransNGramH; // number of non-matched ngrams
        // counters for getting the closest length of reference sentences
        const int TransLen = TransWIdV.Len(); int RefLenSum = 0;
        // go over reference translations and count ngram matches
        TStrV RefTransStrV = TransCorpus->GetRefTransStrV(SentId);
        // we assume that there is at least one reference translation
        IAssert(!RefTransStrV.Empty());
        for (int RefN = 0; RefN < RefTransStrV.Len(); RefN++) {
            // parse reference translation sentence
            TIntV RefWIdV; Parse(RefTransStrV[RefN], RefWIdV);
            TIntH RefNGramH; GetNGramH(RefWIdV, MxNGramLen, RefNGramH);
            // check for matches
            int TransNGramKeyId = TransNGramH.FFirstKeyId();
            while(TransNGramH.FNextKeyId(TransNGramKeyId)) {
                const int NGramId = TransNGramH.GetKey(TransNGramKeyId);
                const int FreeTransNGrams = FreeTransNGramH(NGramId);
                if (RefNGramH.IsKey(NGramId) && (FreeTransNGrams>0)) {
                    // ngram match and still some free ngrams left to clip
                    const int RefNGrams = RefNGramH(NGramId);
                    FreeTransNGramH(NGramId) = TInt::GetMx(0, FreeTransNGrams - RefNGrams);
                }
            }
            // check the length difference
            const int RefLen = RefWIdV.Len();
            RefLenSum += RefLen;
        }
        // count ngrams
        int TransNGramKeyId = TransNGramH.FFirstKeyId();
        while(TransNGramH.FNextKeyId(TransNGramKeyId)) {
            // get ngram
            const int NGramId = TransNGramH.GetKey(TransNGramKeyId);
            IAssert(NGramId != -1);
            // check if two hash tables are aligned (should be...)
            const int FreeNGramId = FreeTransNGramH.GetKey(TransNGramKeyId);
            IAssert(NGramId == FreeNGramId);
            // get ngram count and clip-count
            const int Count = TransNGramH[TransNGramKeyId];
            const int ClipCount = Count - FreeTransNGramH[TransNGramKeyId];
            // add ngram to the coprus ngram counts
            CountNGramH.AddDat(NGramId) += Count;
            ClipCountNGramH.AddDat(NGramId) += ClipCount;
        }
        // count length
        FullTransLen += double(TransLen);
        FullRefLen += double(RefLenSum) / double(RefTransStrV.Len());
    }

    // calculate ngram info scores
    int CountKeyId = CountNGramH.FFirstKeyId();
    while (CountNGramH.FNextKeyId(CountKeyId)) {
        // get the n-gram
        const int NGramId = CountNGramH.GetKey(CountKeyId);
        TIntV NGram = GetNGram(NGramId);
        // prepare counts
        if (NGram.Len() == 1) {
            // n-gram is a word
            const int WordCount = CountNGramH[CountKeyId];
            const double NGramInfoScore = TMath::Log2(FullTransLen / double(WordCount));
            NGramInfoH.AddDat(NGramId, NGramInfoScore);
        } else {
            // more then one word in the n-gram
            // get a n-gram with removed last element
            TIntV N1Gram = NGram; N1Gram.DelLast();
            const int N1GramId = NGramH.GetKeyId(N1Gram);
            // get the counts
            const int NGramCount = CountNGramH(NGramId);
            const int N1GramCount = CountNGramH(N1GramId);
            // get the score
            const double NGramInfoScore = TMath::Log2(double(N1GramCount) / double(NGramCount));
            NGramInfoH.AddDat(NGramId, NGramInfoScore);
        }
    }

    // calcualte ngram precisions
    TFltV ClipCountV(MxNGramLen); ClipCountV.PutAll(0);
    int ClipCountKeyId = ClipCountNGramH.FFirstKeyId();
    while (ClipCountNGramH.FNextKeyId(ClipCountKeyId)) {
        const int NGramId = ClipCountNGramH.GetKey(ClipCountKeyId);
//.........这里部分代码省略.........

void TNetInfBs::GenCascade(TCascade& C, const int& TModel, const double &window, TIntPrIntH& EdgesUsed, const double& delta,
						   const double& std_waiting_time, const double& std_beta) {
	TIntFltH InfectedNIdH; TIntH InfectedBy;
	double GlobalTime; int StartNId;
	double alpha, beta;

	if (GroundTruth->GetNodes() == 0)
		return;

	while (C.Len() < 2) {
		C.Clr();
		InfectedNIdH.Clr();
		InfectedBy.Clr();
		GlobalTime = 0;

		StartNId = GroundTruth->GetRndNId();
		InfectedNIdH.AddDat(StartNId) = GlobalTime;

		while (true) {
			// sort by time & get the oldest node that did not run infection
			InfectedNIdH.SortByDat(true);
			const int& NId = InfectedNIdH.BegI().GetKey();
			GlobalTime = InfectedNIdH.BegI().GetDat();

			// all the nodes has run infection
			if (GlobalTime >= window)
				break;

			// add current oldest node to the network and set its time
			C.Add(NId, GlobalTime);

			// run infection from the current oldest node
			const TNGraph::TNodeI NI = GroundTruth->GetNI(NId);
			for (int e = 0; e < NI.GetOutDeg(); e++) {
				const int DstNId = NI.GetOutNId(e);

				beta = Betas.GetDat(TIntPr(NId, DstNId));

				// flip biased coin (set by beta)
				if (TInt::Rnd.GetUniDev() > beta+std_beta*TFlt::Rnd.GetNrmDev())
					continue;

				alpha = Alphas.GetDat(TIntPr(NId, DstNId));

				// not infecting the parent
				if (InfectedBy.IsKey(NId) && InfectedBy.GetDat(NId).Val == DstNId)
					continue;

				double sigmaT;
				switch (TModel) {
				case 0:
					// exponential with alpha parameter
					sigmaT = TInt::Rnd.GetExpDev(alpha);
					break;
				case 1:
					// power-law with alpha parameter
					sigmaT = TInt::Rnd.GetPowerDev(alpha);
					while (sigmaT < delta) { sigmaT = TInt::Rnd.GetPowerDev(alpha); }
					break;
				case 2:
					// rayleigh with alpha parameter
					sigmaT = TInt::Rnd.GetRayleigh(1/sqrt(alpha));
					break;
				default:
					sigmaT = 1;
					break;
				}

				// avoid negative time diffs in case of noise
				if (std_waiting_time > 0)
					sigmaT = TFlt::GetMx(0.0, sigmaT + std_waiting_time*TFlt::Rnd.GetNrmDev());

				double t1 = GlobalTime + sigmaT;

				if (InfectedNIdH.IsKey(DstNId)) {
					double t2 = InfectedNIdH.GetDat(DstNId);
					if (t2 > t1 && t2 != window) {
						InfectedNIdH.GetDat(DstNId) = t1;
						InfectedBy.GetDat(DstNId) = NId;
					}
				} else {
					InfectedNIdH.AddDat(DstNId) = t1;
					InfectedBy.AddDat(DstNId) = NId;
				}
			}

			// we cannot delete key (otherwise, we cannot sort), so we assign a big time (window cut-off)
			InfectedNIdH.GetDat(NId) = window;
		}

	}

	C.Sort();

	for (TIntH::TIter EI = InfectedBy.BegI(); EI < InfectedBy.EndI(); EI++) {
		TIntPr Edge(EI.GetDat().Val, EI.GetKey().Val);

		if (!EdgesUsed.IsKey(Edge)) EdgesUsed.AddDat(Edge) = 0;

		EdgesUsed.GetDat(Edge) += 1;
//.........这里部分代码省略.........

// Returns the NId where the two directions of search meet up, or -1 if no augmenting path exists. ##TSnap::IntFlowBiDBFS
int IntFlowBiDBFS (const PNEANet &Net, const int& CapIndex, TIntV &Flow, TIntQ &FwdNodeQ, TIntH &PredEdgeH, TIntQ &BwdNodeQ, TIntH &SuccEdgeH, const int& SrcNId, const int& SnkNId) {
  FwdNodeQ.Push(SrcNId);
  PredEdgeH.AddDat(SrcNId, -1);
  BwdNodeQ.Push(SnkNId);
  SuccEdgeH.AddDat(SnkNId, -1);
  while (!FwdNodeQ.Empty() && !BwdNodeQ.Empty()) {
    // Forward search
    const TNEANet::TNodeI &FwdNI = Net->GetNI(FwdNodeQ.Top()); FwdNodeQ.Pop();
    // Check all edges that point into the current node for those over which flow can be returned.
    for (int EdgeN = 0; EdgeN < FwdNI.GetInDeg(); EdgeN++) {
      int NextNId = FwdNI.GetInNId(EdgeN);
      int NextEId = FwdNI.GetInEId(EdgeN);
      if (!PredEdgeH.IsKey(NextNId) && Flow[NextEId] > 0) {
        PredEdgeH.AddDat(NextNId, NextEId);
        if (SuccEdgeH.IsKey(NextNId)) {
          return NextNId;
        }
        FwdNodeQ.Push(NextNId);
      }
    }
    // Check all edges that point out of the current node for those over which flow can be added.
    for (int EdgeN = 0; EdgeN < FwdNI.GetOutDeg(); EdgeN++) {
      int NextNId = FwdNI.GetOutNId(EdgeN);
      int NextEId = FwdNI.GetOutEId(EdgeN);
      if (!PredEdgeH.IsKey(NextNId) && Net->GetIntAttrIndDatE(NextEId, CapIndex) > Flow[NextEId]) {
        PredEdgeH.AddDat(NextNId, NextEId);
        if (SuccEdgeH.IsKey(NextNId)) {
          return NextNId;
        }
        FwdNodeQ.Push(NextNId);
      }
    }
    // Backward search
    const TNEANet::TNodeI &BwdNI = Net->GetNI(BwdNodeQ.Top());  BwdNodeQ.Pop();
    // Check all edges that point out of the current node for those over which flow can be returned.
    for (int EdgeN = 0; EdgeN < BwdNI.GetOutDeg(); EdgeN++) {
      int PrevNId = BwdNI.GetOutNId(EdgeN);
      int PrevEId = BwdNI.GetOutEId(EdgeN);
      if (!SuccEdgeH.IsKey(PrevNId) && Flow[PrevEId] > 0) {
        SuccEdgeH.AddDat(PrevNId, PrevEId);
        if (PredEdgeH.IsKey(PrevNId)) {
          return PrevNId;
        }
        BwdNodeQ.Push(PrevNId);
      }
    }
    // Check all edges that point into the current node for those over which flow can be added.
    for (int EdgeN = 0; EdgeN < BwdNI.GetInDeg(); EdgeN++) {
      int PrevNId = BwdNI.GetInNId(EdgeN);
      int PrevEId = BwdNI.GetInEId(EdgeN);
      if (!SuccEdgeH.IsKey(PrevNId) && Net->GetIntAttrIndDatE(PrevEId, CapIndex) > Flow[PrevEId]) {
        SuccEdgeH.AddDat(PrevNId, PrevEId);
        if (PredEdgeH.IsKey(PrevNId)) {
          return PrevNId;
        }
        BwdNodeQ.Push(PrevNId);
      }
    }
  }
  return -1;
}