from evaluations import parameters
from evaluations.run_experiment import run_experiment
"""
Experiment:
Determine the Influence of the Sybil Node Classifier on Sybilframe
System:
Votetrust Targeted Breadth First
Set of variations:
FN = FP: 0.1, 0.2, 0.4
Yes, large node priors indicate BENIGN labels
"""
evalIntervals = (1,5,10,15,20,25,30,35,40,45,50)
graph = 'slashdot'
for i in (0.1, 0.3, 0.6, 0.8):
paras = parameters.ParameterSettingsP(graph=graph, strategy='breadthFirst', boosted=False, evalAt=evalIntervals, numRepeats=1)
paras.nodeProbSybil = 1-i
paras.nodeProbNonSybil = 0.1
paras.edgeProbSybil = 0.8
paras.edgeProbNonSybil = 0.2
paras.numSeeds = 100
run_experiment(paras, saveAs='./sybilNodeProb/sybilNodeProb{}PTar{}.p'.format(i, graph), systems=('sybilframe',))
from evaluations import parameters
from evaluations.run_experiment import run_experiment
"""
Experiment:
Determine the Influence of the Victim Classifier's performance on the performance of Integro
System:
Votetrust Targeted Breadth First
Set of variations:
d=0.8, 0.9, 0.95, 0.99, 0.99
"""
graph = 'slashdot'
numRepeats = 5
evalInt = (10, 20, 40)
for i in (0.1, 0.2, 0.4):
paras = parameters.ParameterSettingsSR(graph=graph, strategy='breadthFirst', evalAt=evalInt, numRepeats=numRepeats)
paras.nodeProbNonVictim = 0.9
paras.nodeProbVictim = i
run_experiment(paras, saveAs='./victimProb/victimProb{}{}SRTar.p'.format(i, graph), systems=('integro',))
paras = parameters.ParameterSettingsP(graph=graph, strategy='breadthFirst', boosted=False, evalAt=evalInt, numRepeats=numRepeats)
paras.nodeProbNonVictim = 0.9
paras.nodeProbVictim = i
run_experiment(paras, saveAs='./victimProb/victimProb{}{}PTar.p'.format(i, graph), systems=('integro',))
Variations:
Number of seeds: 1, 3, 10, 100, 1000
Systems:
All Systems, all scenarios
"""
for i in (1, 3, 10, 100, 1000):
paras = parameters.ParameterSettingsP(graph='facebook',
strategy='breadthFirst',
boosted=True,
evalAt=(50, ),
numRepeats=3)
paras.numSeeds = i
run_experiment(paras, saveAs='./seeds/seeds{}PTar.p'.format(i))
paras = parameters.ParameterSettingsP(graph='facebook',
strategy='random',
boosted=False,
evalAt=(50, ),
numRepeats=3)
paras.numSeeds = i
run_experiment(paras, saveAs='./seeds/seeds{}PRand.p'.format(i))
paras = parameters.ParameterSettingsSR(graph='facebook',
evalAt=(50, ),
numRepeats=3)
paras.numSeeds = i
run_experiment(paras, saveAs='./seeds/seeds{}SRRand.p'.format(i))
Determine the success rate influences the performance
Variations:
Success Rate:
0.2 - 0.8
0.1 - 0.5
Systems:
All Systems, all scenarios
"""
graph = 'slashdot'
for i in ((0.2, 0.7), (0.1, 0.5)):
paras = parameters.ParameterSettingsP(graph=graph,
strategy='breadthFirst',
boosted="random",
evalAt=(50, ),
numRepeats=3)
paras.acceptanceRatioLimits = i
run_experiment(paras, saveAs='./ratio/ratio{}PTar{}.p'.format(i, graph))
"""
paras = parameters.ParameterSettingsP(graph=graph, strategy='random', boosted=False, evalAt=(50,), numRepeats=3)
paras.acceptanceRatioLimits = i
run_experiment(paras, saveAs='./ratio/ratio{}PRand.p'.format(i))
paras = parameters.ParameterSettingsSR(graph=graph, evalAt=(50,), numRepeats=3)
paras.acceptanceRatioLimits = i
run_experiment(paras, saveAs='./ratio/ratio{}SRRand.p'.format(i))
"""
Votetrust Targeted Breadth First
Set of variations:
FP: 0.1,
FN: 0.1, 0.3, 0.5, 0.6
"""
graph = 'facebook'
for i in (0.1, 0.3, 0.6, 0.81):
for j in (0.1, 0.32, 0.6, 0.8):
paras = parameters.ParameterSettingsP(graph=graph,
strategy='breadthFirst',
boosted=False,
evalAt=(50, ),
numRepeats=3)
paras.edgeProbSybil = 1 - i
paras.edgeProbNonSybil = 0.1
paras.nodeProbSybil = 1 - j
paras.nodeProbNonSybil = 0.1
paras.numSeeds = 100
run_experiment(
paras,
saveAs='./sybilProb/sybilProb{}PTar{}_node{}_edge{}.p'.format(
i, graph,
str(round(1 - paras.nodeProbSybil, 2))[2:],
str(round(1 - paras.edgeProbSybil, 2))[2:]),
systems=('sybilframe', ))
from evaluations import parameters
from evaluations.run_experiment import run_experiment
"""
Experiment:
Determine the Influence of the D parameter on the performance of Votetrust
System:
Votetrust Targeted Breadth First
Set of variations:
d=0.8, 0.9, 0.95, 0.99, 0.999
"""
boosttype = 'random'
graph = 'facebook'
for i in (0.8, 0.99, 0.999):
paras = parameters.ParameterSettingsP(graph=graph,
strategy='breadthFirst',
boosted=boosttype,
evalAt=(50, ),
numRepeats=5)
paras.d = i
paras.numSeeds = 100
run_experiment(paras,
saveAs='./d/d{}PTar_{}_{}.p'.format(i, boosttype, graph),
systems=('votetrust', ))