def classifyTop500(self):
resourceIds = self.getAvailableResourceIds()
classificator = Classificator()
for resourceId in resourceIds[:500]:
print "processing %s" % resourceId
try:
classified = classificator.classifyResource(resourceId)
pickle.dump(classified, open(data_classified_path + resourceId, 'wb'))
except BaseException as e:
print str(e)
def classifyTheSame(self):
resourceIds = self.getClassifiedResourceIds()
classificator = Classificator(4) #number is for foxlight
for resourceId in resourceIds:
print "processing %s" % resourceId
try:
classified = classificator.classifyResource(resourceId, classifierName="Spotlight")
pickle.dump(classified, open(data_classified_path + 'spotlight/' + resourceId, 'wb'))
except BaseException as e:
print str(e)
def getEntitiesTop10(self):
resourceIds = self.getAvailableResourceIds()
classificator = Classificator()
for resourceId in resourceIds[3:10]:
print "processing %s" % resourceId
try:
classified = classificator.getEntities(resourceId)
pickle.dump(classified, open(data_classified_path + 'entities/' + resourceId, 'wb'))
except BaseException as e:
print str(e)
break
