// 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);
}
// actors collaboration graph
PUNGraph TImdbNet::GetActorGraph() const {
TIntPrSet EdgeSet;
for (TNodeI NI = BegNI(); NI < EndNI(); NI++) {
if (NI().GetTy() == mtyActor) {
const int NId1 = NI.GetId();
for (int e = 0; e < NI.GetOutDeg(); e++) {
if (NI.GetOutNDat(e).GetTy() != mtyActor) {
TNodeI NI2 = GetNI(NI.GetOutNId(e));
for (int e2 = 0; e2 < NI2.GetInDeg(); e2++) {
if (NI2.GetInNDat(e2).GetTy() == mtyActor) {
const int NId2 = NI2.GetInNId(e2);
EdgeSet.AddKey(TIntPr(TMath::Mn(NId1, NId2), TMath::Mx(NId1, NId2)));
}
}
}
}
}
}
PUNGraph G = TUNGraph::New();
for (int i = 0; i < EdgeSet.Len(); i++) {
const int NId1 = EdgeSet[i].Val1;
const int NId2 = EdgeSet[i].Val2;
if (! G->IsNode(NId1)) { G->AddNode(NId1); }
if (! G->IsNode(NId2)) { G->AddNode(NId2); }
G->AddEdge(NId1, NId2);
}
return G;
}
// 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);
}