def ccScores(self):
data=self.data
columnNames=self.columnNames
totalSize=self.totalSize
relationships=self.relationships
size=len(data)
bitmap={}
for i,column in enumerate(columnNames):
relationships[column]={}
bitmap[column]={} #this is a dictionary which is a set of flags per data value remembering if the increment already happened.
for element in data:
item=element[i]
rlist=sparqlQuerypy.findBottomUp(item.strip())
print 'number of nodes for', item.strip(), " is ", len(rlist)
bitmap[column][item]={}
for r in rlist:
if r[0] not in bitmap[column][item].keys():
bitmap[column][item][r[0]]=0
if r[0] not in relationships[column].keys():
relationships[column][r[0]]={}
relationships[column][r[0]]['name']='cc'
if 'incoming' not in relationships[column][r[0]].keys():
relationships[column][r[0]]['incoming']=1
relationships[column][r[0]]['cc']=1.0/totalSize
else:
relationships[column][r[0]]['incoming']+=1
relationships[column][r[0]]['cc']=relationships[column][r[0]]['incoming']*1.0/totalSize
bitmap[column][item][r[0]]=1
classSet=set() # A set to save all the possible cc classes for ease of retrieval later and to streamline it.
for column in columnNames: #Loop to push the relations and nodes to Neo4j
for classes in relationships[column].keys():
classSet.add(classes)
rel_data=Neo4jDrive.insertNodeAndRelationship(column,'cc',classes)[0]
rel_data.properties['rel_class']='cc'
rel_data.properties['fk']=relationships[column][classes]['cc']
rel_data.push()
for classes in classSet: #Loop to update the CCS score for each class after the previous loop is over. CCS=sum(fk)/no(fk) for the node.
print classes
cummulative=0 # The accumulator
linkNumbers=0 # The denominator
for link in Neo4jDrive.findIncomingCCLinks(classes): #loop to find incoming cc edges.
cummulative+=link[0].properties['fk']
linkNumbers+=1
node=Neo4jDrive.findNodeByName(classes)
node.properties['ccs']=cummulative*1.0/linkNumbers
node.properties['type']='cc'
node.push()
def run(self):
support=self.support
totalNumberOfValues=self.totalNumberOfValues*1.0
column=self.column
columnNames=self.columnNames
item=self.item
rlist=sparqlQuerypy.findBottomUp(item.strip())
print 'number of nodes for', item.strip(), " is ", len(rlist)
log.write('number of nodes for'+str( item.strip())+ " is "+ str(len(rlist))+'\n')
flag=0
for r in rlist:
rel_data=Neo4jDrive.insertNodeAndRelationship(columnNames[column],"cc",r[2])
rel_data=rel_data[0]
node=Neo4jDrive.findNodeByName(r[2])
if r[2]=='http://dbpedia.org/ontology/PopulatedPlace':
print columnNames[column], 'Happening'
print 'potato',rel_data
if rel_data.properties['incoming']==None: #find out why this is not happenings
rel_data.properties['incoming']=1
rel_data.properties['ccs']=1/totalNumberOfValues
rel_data.push()
#print 'tomato',rel_data
else:
if flag==0:
rel_data.properties['incoming']+=1
rel_data.push()
rel_data.properties['ccs']=node.properties['incoming']/totalNumberOfValues
flag=1
node.properties['type']='cc'
node.properties['ccs']=0
numberOfLinks=0
for link in Neo4jDrive.findIncomingCCLinks(r[2]):
node.properties['ccs']+=link[0].properties['ccs']
numberOfLinks+=1
if numberOfLinks>0: node.properties['ccs']/=numberOfLinks
node.push()
rel_data.properties['rel_class'] = 'cc'
#rel_data.properties['ccs']=node.proper/(totalNumberOfValues*1.0)
rel_data.push()
