int main(int argc, char* argv[]) {
Env = TEnv(argc, argv, TNotify::StdNotify);
Env.PrepArgs(TStr::Fmt("cesna. build: %s, %s. Time: %s", __TIME__, __DATE__, TExeTm::GetCurTm()));
TExeTm ExeTm;
Try
TStr OutFPrx = Env.GetIfArgPrefixStr("-o:", "", "Output Graph data prefix");
const TStr InFNm = Env.GetIfArgPrefixStr("-i:", "./1912.edges", "Input edgelist file name");
const TStr LabelFNm = Env.GetIfArgPrefixStr("-l:", "", "Input file name for node names (Node ID, Node label) ");
const TStr AttrFNm = Env.GetIfArgPrefixStr("-a:", "./1912.nodefeat", "Input node attribute file name");
const TStr ANameFNm = Env.GetIfArgPrefixStr("-n:", "./1912.nodefeatnames", "Input file name for node attribute names");
int OptComs = Env.GetIfArgPrefixInt("-c:", 10, "The number of communities to detect (-1: detect automatically)");
const int MinComs = Env.GetIfArgPrefixInt("-mc:", 3, "Minimum number of communities to try");
const int MaxComs = Env.GetIfArgPrefixInt("-xc:", 20, "Maximum number of communities to try");
const int DivComs = Env.GetIfArgPrefixInt("-nc:", 5, "How many trials for the number of communities");
const int NumThreads = Env.GetIfArgPrefixInt("-nt:", 4, "Number of threads for parallelization");
const double AttrWeight = Env.GetIfArgPrefixFlt("-aw:", 0.5, "We maximize (1 - aw) P(Network) + aw * P(Attributes)");
const double LassoWeight = Env.GetIfArgPrefixFlt("-lw:", 1.0, "Weight for l-1 regularization on learning the logistic model parameters");
const double StepAlpha = Env.GetIfArgPrefixFlt("-sa:", 0.05, "Alpha for backtracking line search");
const double StepBeta = Env.GetIfArgPrefixFlt("-sb:", 0.3, "Beta for backtracking line search");
const double MinFeatFrac = Env.GetIfArgPrefixFlt("-mf:", 0.0, "If the fraction of nodes with positive values for an attribute is smaller than this, we ignore that attribute");
#ifdef USE_OPENMP
omp_set_num_threads(NumThreads);
#endif
PUNGraph G;
TIntStrH NIDNameH;
TStrHash<TInt> NodeNameH;
TVec<TFltV> Wck;
TVec<TIntV> EstCmtyVV;
if (InFNm.IsStrIn(".ungraph")) {
TFIn GFIn(InFNm);
G = TUNGraph::Load(GFIn);
} else {
G = TAGMUtil::LoadEdgeListStr<PUNGraph>(InFNm, NodeNameH);
NIDNameH.Gen(NodeNameH.Len());
for (int s = 0; s < NodeNameH.Len(); s++) { NIDNameH.AddDat(s, NodeNameH.GetKey(s)); }
}
if (LabelFNm.Len() > 0) {
TSsParser Ss(LabelFNm, ssfTabSep);
while (Ss.Next()) {
if (Ss.Len() > 0) { NIDNameH.AddDat(Ss.GetInt(0), Ss.GetFld(1)); }
}
}
printf("Graph: %d Nodes %d Edges\n", G->GetNodes(), G->GetEdges());
//load attribute
TIntV NIDV;
G->GetNIdV(NIDV);
THash<TInt, TIntV> RawNIDAttrH, NIDAttrH;
TIntStrH RawFeatNameH, FeatNameH;
if (ANameFNm.Len() > 0) {
TSsParser Ss(ANameFNm, ssfTabSep);
while (Ss.Next()) {
if (Ss.Len() > 0) { RawFeatNameH.AddDat(Ss.GetInt(0), Ss.GetFld(1)); }
}
}
TCesnaUtil::LoadNIDAttrHFromNIDKH(NIDV, AttrFNm, RawNIDAttrH, NodeNameH);
TCesnaUtil::FilterLowEntropy(RawNIDAttrH, NIDAttrH, RawFeatNameH, FeatNameH, MinFeatFrac);
TExeTm RunTm;
TCesna CS(G, NIDAttrH, 10, 10);
if (OptComs == -1) {
printf("finding number of communities\n");
OptComs = CS.FindComs(NumThreads, MaxComs, MinComs, DivComs, "", false, 0.1, StepAlpha, StepBeta);
}
CS.NeighborComInit(OptComs);
CS.SetWeightAttr(AttrWeight);
CS.SetLassoCoef(LassoWeight);
if (NumThreads == 1 || G->GetEdges() < 1000) {
CS.MLEGradAscent(0.0001, 1000 * G->GetNodes(), "", StepAlpha, StepBeta);
} else {
CS.MLEGradAscentParallel(0.0001, 1000, NumThreads, "", StepAlpha, StepBeta);
}
CS.GetCmtyVV(EstCmtyVV, Wck);
TAGMUtil::DumpCmtyVV(OutFPrx + "cmtyvv.txt", EstCmtyVV, NIDNameH);
FILE* F = fopen((OutFPrx + "weights.txt").CStr(), "wt");
if (FeatNameH.Len() == Wck[0].Len()) {
fprintf(F, "#");
for (int k = 0; k < FeatNameH.Len(); k++) {
fprintf(F, "%s", FeatNameH[k].CStr());
if (k < FeatNameH.Len() - 1) { fprintf(F, "\t"); }
}
fprintf(F, "\n");
}
for (int c = 0; c < Wck.Len(); c++) {
for (int k = 0; k < Wck[c].Len(); k++) {
fprintf(F, "%f", Wck[c][k].Val);
if (k < Wck[c].Len() - 1) { fprintf(F, "\t"); }
}
fprintf(F, "\n");
}
fclose(F);
Catch
printf("\nrun time: %s (%s)\n", ExeTm.GetTmStr(), TSecTm::GetCurTm().GetTmStr().CStr());
return 0;
}
