void TCliqueOverlap::GetIntersection(const THashSet<TInt>& A, const THashSet<TInt>& B, THashSet<TInt>& C) {
if (A.Len() < B.Len()) {
for (THashSetKeyI<TInt> it=A.BegI(); it<A.EndI(); it++)
if (B.IsKey(it.GetKey())) C.AddKey(it.GetKey());
} else {
for (THashSetKeyI<TInt> it=B.BegI(); it<B.EndI(); it++)
if (A.IsKey(it.GetKey())) C.AddKey(it.GetKey());
}
}
int TCliqueOverlap::Intersection(const THashSet<TInt>& A, const THashSet<TInt>& B) {
int n = 0;
if (A.Len() < B.Len()) {
for (THashSetKeyI<TInt> it=A.BegI(); it<A.EndI(); it++)
if (B.IsKey(it.GetKey())) n++;
} else {
for (THashSetKeyI<TInt> it=B.BegI(); it<B.EndI(); it++)
if (A.IsKey(it.GetKey())) n++;
}
return n;
}
/// Shingles by words
void LSH::HashShinglesOfClusters(TQuoteBase *QuoteBase,
TClusterBase *ClusterBase, TIntV& ClusterIds, TInt ShingleLen,
THash<TMd5Sig, TIntV>& ShingleToClusterIds) {
Err("Hashing shingles of clusters...\n");
for (int i = 0; i < ClusterIds.Len(); i++) {
if (i % 1000 == 0) {
fprintf(stderr, "%d out of %d completed\n", i, ClusterIds.Len());
}
TCluster C;
ClusterBase->GetCluster(ClusterIds[i], C);
//fprintf(stderr, "%d vs. %d\n", ClusterIds[i].Val, C.GetId().Val);
// Put x-word shingles into hash table; x is specified by ShingleLen parameter
THashSet < TMd5Sig > CHashedShingles;
GetHashedShinglesOfCluster(QuoteBase, C, ShingleLen, CHashedShingles);
for (THashSet<TMd5Sig>::TIter Hash = CHashedShingles.BegI();
Hash < CHashedShingles.EndI(); Hash++) {
TIntV ShingleClusterIds;
if (ShingleToClusterIds.IsKey(*Hash)) {
ShingleClusterIds = ShingleToClusterIds.GetDat(*Hash);
}
ShingleClusterIds.Add(ClusterIds[i]);
ShingleToClusterIds.AddDat(*Hash, ShingleClusterIds);
}
}
Err("Done hashing!\n");
}
int TCliqueOverlap::GetNodeIdWithMaxDeg(const THashSet<TInt>& Set) const{
int id = -1;
int maxDeg = -1;
//
for (THashSetKeyI<TInt> it=Set.BegI(); it<Set.EndI(); it++) {
int nId = it.GetKey();
int deg = m_G->GetNI(nId).GetDeg();
if (maxDeg < deg) { maxDeg=deg; id=nId; }
}
return id;
}
int main(int argc, char *argv[]) {
TStr BaseString = "/lfs/1/tmp/curis/week/QBDB.bin";
TFIn BaseFile(BaseString);
TQuoteBase *QB = new TQuoteBase;
TDocBase *DB = new TDocBase;
QB->Load(BaseFile);
DB->Load(BaseFile);
TIntV QuoteIds;
QB->GetAllQuoteIds(QuoteIds);
int NumQuotes = QuoteIds.Len();
THash<TInt, TStrSet> PeakCounts;
for (int i = 0; i < NumQuotes; i++) {
TQuote CurQuote;
if (QB->GetQuote(QuoteIds[i], CurQuote)) {
TVec<TSecTm> Peaks;
CurQuote.GetPeaks(DB, Peaks);
TStr QuoteString;
CurQuote.GetParsedContentString(QuoteString);
TStrSet StringSet;
if (PeakCounts.IsKey(Peaks.Len())) {
StringSet = PeakCounts.GetDat(Peaks.Len());
}
StringSet.AddKey(QuoteString);
PeakCounts.AddDat(Peaks.Len(), StringSet);
}
}
TIntV PeakCountKeys;
PeakCounts.GetKeyV(PeakCountKeys);
PeakCountKeys.Sort(true);
for (int i = 0; i < PeakCountKeys.Len(); i++) {
TStrSet CurSet = PeakCounts.GetDat(PeakCountKeys[i]);
if (CurSet.Len() > 0) {
printf("QUOTES WITH %d PEAKS\n", PeakCountKeys[i].Val);
printf("#########################################\n");
THashSet<TStr> StringSet = PeakCounts.GetDat(PeakCountKeys[i]);
for (THashSet<TStr>::TIter l = StringSet.BegI(); l < StringSet.EndI(); l++) {
printf("%s\n", l.GetKey().CStr());
}
printf("\n");
}
}
delete QB;
delete DB;
return 0;
}
int TCliqueOverlap::MaxNbrsInCANDNodeId(const THashSet<TInt>& SUBG, const THashSet<TInt>& CAND) const{
int id = -1;
int maxIntersection = -1;
//
for (THashSetKeyI<TInt> it=SUBG.BegI(); it<SUBG.EndI(); it++) {
int nId = it.GetKey();
TUNGraph::TNodeI nIt = m_G->GetNI(nId);
int deg = nIt.GetDeg();
//
int curIntersection = 0;
for (int i=0; i<deg; i++) {
int nbrId = nIt.GetNbrNId(i);
if (CAND.IsKey(nbrId)) curIntersection++;
}
//
if (maxIntersection < curIntersection) { maxIntersection=curIntersection; id=nId; }
}
return id;
}
/////////////////////////////////////////////////
// TCommunity implementation
void TCliqueOverlap::GetRelativeComplement(const THashSet<TInt>& A, const THashSet<TInt>& B, THashSet<TInt>& Complement) {
for (THashSet<TInt>::TIter it=A.BegI(); it<A.EndI(); it++) {
const int nId = it.GetKey();
if (!B.IsKey(nId)) Complement.AddKey(nId);
}
}
