def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.QExpInName = conf.GetConf('qexp')
self.lClusterLmInName = conf.GetConf('clusterlm')
if type(self.lClusterLmInName) != list:
self.lClusterLmInName = [self.lClusterLmInName]
self.OutName = conf.GetConf('out')
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.CtfCenter.Load(conf.GetConf('termctf'))
self.Word2VecFile = conf.GetConf('word2vec')
self.CateDenseCenter.load(conf.GetConf('cateattdense'))
self.FbObjCacheCenter.SetConf(ConfIn)
print "inited"
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.WorkDir = conf.GetConf('objcachedir') + '/'
# print "set obj cachedir to [%s]" %(self.WorkDir)
# self.WriteCache = bool(int(conf.GetConf('writecache',self.WriteCache)))
# self.CreateHash()
return True
def SetConf(self, ConfIn):
conf = cxConf(ConfIn)
self.QExpInName = conf.GetConf('qexp')
self.lClusterLmInName = conf.GetConf('clusterlm')
if type(self.lClusterLmInName) != list:
self.lClusterLmInName = [self.lClusterLmInName]
self.OutName = conf.GetConf('out')
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.CashDir = conf.GetConf('cashdir')
self.QueryIn = conf.GetConf('in')
self.FaccDir = conf.GetConf('faccdir')
self.OutDir = conf.GetConf('outdir')
self.SERPDepth = int(conf.GetConf('serpdepth',self.SERPDepth))
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
# self.WOrig = float(conf.LoadConf("worig"))
self.DirMu = float(conf.GetConf('dirmu',self.DirMu))
self.NumOfExpTerm = int(conf.GetConf("numofexpterm",self.NumOfExpTerm))
self.PrfDocNum = int(conf.GetConf('prfdocnum',self.PrfDocNum))
self.CtfCenter.Load(conf.GetConf("ctfpath"))
return True
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.FilterCotype = bool(conf.GetConf('cotypefilter',self.CotypeFilter))
self.filterdomain = conf.GetConf('filterdomain')
if type(self.filterdomain) != list:
self.filterdomain = [self.filterdomain]
self.InGraphDir = conf.GetConf('ingraphdir')
self.OutGraphDir = conf.GetConf('outgraphdir')
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.NumOfObjUsed = int(conf.GetConf('numofobj',self.NumOfObjUsed))
self.CtfCenter.Load(conf.GetConf('ctfpath'))
self.ObjCenter.SetConf(ConfIn)
self.NumOfExpTerm = int(conf.GetConf('numofexpterm',self.NumOfExpTerm))
self.QObjRankInName = conf.GetConf('queryobjrank')
return
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.InName = conf.GetConf('in')
self.OutName = conf.GetConf('out')
self.k = int(conf.GetConf('k',self.k))
self.MidDir = conf.GetConf('middir')
self.Model = cxKMeansC()
self.Model.SetK(self.k)
self.Model.SetWorkDir(self.MidDir)
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.lEdgeFeature = conf.GetConf('edgefeature', self.lEdgeFeature)
self.KeepPRFFeature = bool(int(conf.GetConf('keepprf',1)))
self.KeepPRALvl1Feature = bool(int(conf.GetConf('keeplvl1',1)))
self.KeepPRALvl2Feature = bool(int(conf.GetConf('keeplvl2',0)))
self.OutName = conf.GetConf('out')
self.TermIn = conf.GetConf('in')
self.EdgeTypeGrouping = conf.GetConf('edgetypegrouping',self.EdgeTypeGrouping)
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.FaccDocDir= conf.GetConf('faccdocdir')
self.InName = conf.GetConf('in')
self.OutName = conf.GetConf('out')
CtfInName = conf.GetConf('objctf')
if "" != CtfInName:
self.FaccCtf.Load(CtfInName)
print "ctf load from [%s]" %(CtfInName)
def SetConf(self,ConfIn):
self.LinearReranker.SetConf(ConfIn)
self.Evaluator.SetConf(ConfIn)
conf = cxConf(ConfIn)
self.IndriResDir = conf.GetConf('cashdir')
self.NewTermW = float(conf.GetConf('newtermweight',self.NewTermW))
self.EvaDepth = self.Evaluator.Depth
self.UseBinaryScore = bool(int(conf.GetConf('usebinaryscore',0)))
def ObjVecMakerUnitRun(ConfIn):
conf = cxConf(ConfIn)
InName = conf.GetConf('in')
OutName = conf.GetConf('out')
Maker = ObjVecMakerC(ConfIn)
Maker.ProcessQObjFile(InName, OutName)
return True
def SetConf(self, ConfIn):
conf = cxConf(ConfIn)
self.InName = conf.GetConf('in')
self.OutName = conf.GetConf('out')
self.k = int(conf.GetConf('k', self.k))
self.MidDir = conf.GetConf('middir')
self.Model = cxKMeansC()
self.Model.SetK(self.k)
self.Model.SetWorkDir(self.MidDir)
def SetConf(self, ConfIn):
conf = cxConf(ConfIn)
self.NumOfObjUsed = int(conf.GetConf('numofobj', self.NumOfObjUsed))
self.CtfCenter.Load(conf.GetConf('ctfpath'))
self.ObjCenter.SetConf(ConfIn)
self.NumOfExpTerm = int(conf.GetConf('numofexpterm',
self.NumOfExpTerm))
self.QObjRankInName = conf.GetConf('queryobjrank')
return
def BfsQueryFreebaseUnitRun(ConfIn):
conf = cxConf(ConfIn)
InName = conf.GetConf('in')
BFSer = BfsQueryFreebaseC(ConfIn)
for line in open(InName):
qid,query = line.strip().split('\t')
BFSer.BFS(qid,query)
BFSer.CleanUp()
return True
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.OutDir = conf.GetConf('out')
if not os.path.isdir(self.OutDir):
os.makedirs(self.OutDir)
self.EdgeObjIn = conf.GetConf('in')
self.FbDumpIn = conf.GetConf('fbdumpin')
self.FaccDir = conf.GetConf('faccdir')
self.MaxOccurPerEdge = int(conf.GetConf('maxoccurperedge',self.MaxOccurPerEdge))
def QuerySubgraphConstructorUnitRun(ConfIn):
conf = cxConf(ConfIn)
InName = conf.GetConf('in')
BFSer = QuerySubgraphConstructorC(ConfIn)
for line in open(InName):
qid,query = line.strip().split('\t')
BFSer.BFS(qid,query)
BFSer.CleanUp()
return True
def QExpParaEvaResCollectorUnitTest(ConfIn):
QExpParaEvaResCollectorC.ShowConf()
print "out"
conf = cxConf(ConfIn)
OutName = conf.GetConf('out')
Collector = QExpParaEvaResCollectorC(ConfIn)
hDict = Collector.Process()
out = open(OutName,'w')
json.dump(hDict,out)
out.close()
print "done"
return True
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.WorkDir = conf.GetConf('workdir')
self.ObjectCashDir = self.WorkDir + "/obj"
if not os.path.isdir(self.ObjectCashDir):
os.makedirs(self.ObjectCashDir)
self.BFSLvl = int(conf.GetConf('bfslvl',self.BFSLvl))
self.MaxSearchRes = int(conf.GetConf('maxsearchres',self.MaxSearchRes))
self.MaxCoTypeExp = int(conf.GetConf('maxcotypeexp',self.MaxCoTypeExp))
self.MaxNeighborExp = int(conf.GetConf('maxneighborexp',self.MaxNeighborExp))
self.DumpCache = bool(int(conf.GetConf('dumpcache',1)))
self.load()
return True
def SetConf(self,ConfIn):
self.CandidateTermGetter.SetConf(ConfIn)
self.TermLabelGetter.SetConf(ConfIn)
self.FeatureExtractor.SetConf(ConfIn)
conf = cxConf(ConfIn)
self.OutFormat = conf.GetConf('outformat')
self.TotalDocToRead = conf.GetConf('totalserpnum')
self.CashDir = conf.GetConf('cashdir')
self.GenerateTerm = bool(int(conf.GetConf('generateterm',1)))
self.LabelTerm = bool(int(conf.GetConf('labelterm',1)))
self.ExtractPRFFeature = bool(int(conf.GetConf('extractprffeature',1)))
return True
def SingleTermPerformanceUnitTest(ConfIn):
conf = cxConf(ConfIn)
InName = conf.GetConf('in')
OutName = conf.GetConf('out')
SingleTermPerformance = SingleTermPerformanceC(ConfIn)
CurrentQ = ""
CurrentQid = ""
lTerm = []
out = open(OutName,'w')
for line in open(InName):
line = line.strip()
vCol = line.split('\t')
if len(vCol) < 3:
continue
qid = vCol[0]
query = vCol[1]
term = vCol[2]
if CurrentQ == "":
CurrentQ = query
CurrentQid = qid
if CurrentQ != query:
print "working query [%s]" %(CurrentQ)
lTermScore = SingleTermPerformance.EvaluatePerQ(CurrentQid, CurrentQ, lTerm)
for i in range(len(lTerm)):
print >>out, qid + '\t' + query + '\t' + lTerm[i] + '\t%f' %(lTermScore[i])
CurrentQ = query
CurrentQid = qid
lTerm = []
lTerm.append(term)
print "working query [%s]" %(CurrentQ)
lTermScore = SingleTermPerformance.EvaluatePerQ(CurrentQid, CurrentQ, lTerm)
for i in range(len(lTerm)):
print >>out, qid + '\t' + query + '\t' + lTerm[i] + '\t%f' %(lTermScore[i])
out.close()
return True
global MinGroup
global MaxGroup
llSubGroup = []
for i in range(MinGroup,min(MaxGroup,len(lGroup))+1):
lSubGroup = [list(item) for item in itertools.combinations(lGroup,i)]
llSubGroup.extend(lSubGroup)
return llSubGroup
if 2 != (len(sys.argv)):
print "conf:\nin\noutdir\ngroup prf#pralvl0#pralvl1#word2vec#hyper\nmingroup\nmaxgroup"
sys.exit()
conf = cxConf(sys.argv[1])
InName = conf.GetConf('in')
OutDir = conf.GetConf('outdir')
if not os.path.isdir(OutDir):
os.makedirs(OutDir)
lGroup = conf.GetConf('group')
MinGroup = int(conf.GetConf('mingroup'))
MaxGroup = int(conf.GetConf('maxgroup'))
llSubGroup = EnumrateGroup(lGroup)
llExpTerm = ReadQExpTerms(InName)
for lSubGroup in llSubGroup:
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.FbDumpIn = conf.GetConf('dumpin')
self.OutName = conf.GetConf('out')
def SetConf(self, ConfIn):
conf = cxConf(ConfIn)
self.FbDumpIn = conf.GetConf('dumpin')
self.OutName = conf.GetConf('out')
def SetConf(self,ConfIn):
super(QExpParaEvaResCollectorC,self).SetConf(ConfIn)
conf = cxConf(ConfIn)
self.MainEvaMethod = conf.GetConf('mainevamethod')
return True
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.InName = conf.GetConf('in')
self.OutName = conf.GetConf('out')
self.Model.SetWorkDir(conf.GetConf('middir'))
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.InQuery = conf.GetConf('in')
self.OutName = conf.GetConf('out')
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.FaccDir = conf.GetConf('faccdir')
self.InTargetObj = conf.GetConf('in')
self.OutName = conf.GetConf('out')
self.InType = conf.GetConf('fakba')
def SetConf(self, ConfIn):
conf = cxConf(ConfIn)
self.InName = conf.GetConf('in')
self.OutName = conf.GetConf('out')
self.Model.SetWorkDir(conf.GetConf('middir'))
def SetConf(self,ConfIn):
super(QuerySubgraphTargetTermFilterC,self).SetConf(ConfIn)
conf = cxConf(ConfIn)
self.GraphDumpDir = conf.GetConf('graphdumpdir')
self.OutDir = conf.GetConf('outdir')
hResExp = {}
if not Intersect:
hResExp = deepcopy(hExpTerm)
else:
for item in hIntersect:
hResExp[item] = deepcopy(hExpTerm[item])
return hResExp
if 2 != len(sys.argv):
print "conf:"
print "target\nmergefile\nintersect 1\nout"
sys.exit()
conf = cxConf(sys.argv[1])
TargetIn = conf.GetConf('target')
lInName = conf.GetConf('mergefile')
OutName = conf.GetConf('out')
if type(lInName) == str:
lInName = [lInName]
Intersect = bool(int(conf.GetConf('intersect', 1)))
hExpTerm = LoadAndMergeOneExpTerm(TargetIn)
hResExp = MergeFilesToADict(hExpTerm, lInName, Intersect)
lExpTerm = [ExpTerm for key, ExpTerm in hResExp.items()]
lExpTerm.sort(key=attrgetter('qid'))
out = open(OutName, 'w')
for ExpTerm in lExpTerm:
def SetConf(self,ConfIn):
conf = cxConf(ConfIn)
self.k = int(conf.GetConf('k', self.k))
self.InName = conf.GetConf('in')
self.OutName = conf.GetConf('out')
