def step05(paramFile):
    #util = ParameterUtil(parameter_file = 'data/formatado/arxiv/nowell_example_1994_1999.txt')
    util = ParameterUtil(parameter_file=paramFile)

    myparams = Parameterization(util.keyword_decay, util.lengthVertex, util.t0,
                                util.t0_, util.t1, util.t1_,
                                util.FeaturesChoiced, util.graph_file,
                                util.trainnig_graph_file, util.test_graph_file,
                                util.decay)
    calc = Calculate(myparams, util.nodes_notlinked_file, util.calculated_file,
                     util.ordered_file, util.maxmincalculated_file)
    myparams.generating_Test_Graph()
    analise = Analyse(
        myparams, FormatingDataSets.get_abs_file_path(util.calculated_file),
        FormatingDataSets.get_abs_file_path(util.analysed_file) +
        '.random.analised.txt', calc.qtyDataCalculated)
示例#2
0
def step07(paramFile):
    #util = ParameterUtil(parameter_file = 'data/formatado/arxiv/nowell_example_1994_1999.txt')
    util = ParameterUtil(parameter_file=paramFile)

    myparams = Parameterization(util.keyword_decay, util.lengthVertex, util.t0,
                                util.t0_, util.t1, util.t1_,
                                util.FeaturesChoiced, util.graph_file,
                                util.trainnig_graph_file, util.test_graph_file,
                                util.decay)
    calc = Calculate(myparams, util.nodes_notlinked_file, util.calculated_file,
                     util.ordered_file, util.maxmincalculated_file)
    myparams.generating_Test_Graph()
    topRank = Analyse.getTopRank(util.analysed_file + '.random.analised.txt')
    print 'Analising Files with TopRank', str(topRank)
    for OrderingFilePath in calc.getfilePathOrdered_separeted():
        analise = Analyse(myparams, OrderingFilePath,
                          OrderingFilePath + '.analised.txt', topRank)
示例#3
0
if __name__ == '__main__':
    util = ParameterUtil(parameter_file = 'data/formatado/exemplomenor/config/config.txt')
    myparams = Parameterization(t0 = util.t0, t0_ = util.t0_, t1 = util.t1, t1_ = util.t1_, 
                                filePathGraph = util.graph_file, filePathTrainingGraph = util.trainnig_graph_file, filePathTestGraph = util.test_graph_file, decay = util.decay, domain_decay = util.domain_decay, min_edges = util.min_edges, scoreChoiced = util.ScoresChoiced, weightsChoiced = util.WeightsChoiced, weightedScoresChoiced = util.WeightedScoresChoiced, FullGraph = None)

    myparams.generating_Training_Graph()
    myparams.generating_Test_Graph()
    
    selection = VariableSelection(myparams.trainnigGraph, util.min_edges)
    nodesNotLinked = selection.get_pair_nodes_not_linked()
    calc = CalculateInMemory(myparams, nodesNotLinked)
    resultsCalculate = calc.executingCalculate()
    
    
    calc.Separating_calculateFile()
    analise = Analyse(myparams, FormatingDataSets.get_abs_file_path(util.calculated_file), FormatingDataSets.get_abs_file_path(util.analysed_file) + '.random.analised.txt', calc.qtyDataCalculated)
    topRank = Analyse.getTopRank(util.analysed_file + '.random.analised.txt')
    calc.Ordering_separating_File(topRank)
    for OrderingFilePath in calc.getfilePathOrdered_separeted():
        analise = Analyse(myparams, OrderingFilePath, OrderingFilePath + '.analised.txt', topRank )
    
    
    print "Trainning Period:", myparams.t0, " - ", myparams.t0_
    print "Test Period:", myparams.t1, " - ", myparams.t1_
    
    print "# Papers in Trainning: ",  myparams.get_edges(myparams.trainnigGraph)
    print "# Authors in Training: ", myparams.get_nodes(myparams.trainnigGraph)
    print "# Papers in Test: ",  myparams.get_edges(myparams.testGraph)
    print "# Authors in Test", myparams.get_nodes(myparams.testGraph)
    
    print "# pair of Authors with at least 3 articles Calculated: ", calc.qtyDataCalculated  #FormatingDataSets.getTotalLineNumbers(FormatingDataSets.get_abs_file_path(util.calculated_file))
示例#4
0
'''
Created on Aug 22, 2015

@author: cptullio
Generating TopRank
'''
from parametering.ParameterUtil import ParameterUtil
from parametering.Parameterization import Parameterization
from calculating.Calculate import Calculate
from analysing.Analyse import Analyse
from formating.FormatingDataSets import FormatingDataSets

if __name__ == '__main__':

    util = ParameterUtil(
        parameter_file='data/formatado/duarte/nowell_duarte_1994_1999.txt')
    myparams = Parameterization(util.keyword_decay, util.lengthVertex, util.t0,
                                util.t0_, util.t1, util.t1_,
                                util.FeaturesChoiced, util.graph_file,
                                util.trainnig_graph_file, util.test_graph_file,
                                util.decay)
    calc = Calculate(myparams, util.nodes_notlinked_file, util.calculated_file,
                     util.ordered_file, util.maxmincalculated_file)
    myparams.generating_Test_Graph()
    analise = Analyse(
        myparams, FormatingDataSets.get_abs_file_path(util.calculated_file),
        FormatingDataSets.get_abs_file_path(util.analysed_file) +
        '.random.analised.txt', calc.qtyDataCalculated)
示例#5
0
'''
Created on Aug 22, 2015

@author: cptullio
Analysing the results
'''
from parametering.ParameterUtil import ParameterUtil
from parametering.Parameterization import Parameterization
from calculating.Calculate import Calculate
from analysing.Analyse import Analyse

if __name__ == '__main__':
    util = ParameterUtil(parameter_file = 'data/formatado/duarte/nowell_duarte_1994_1999.txt')
    myparams = Parameterization(util.keyword_decay, util.lengthVertex, util.t0, util.t0_, util.t1, util.t1_, util.FeaturesChoiced, util.graph_file, util.trainnig_graph_file, util.test_graph_file, util.decay)
    calc = Calculate(myparams, util.nodes_notlinked_file, util.calculated_file, util.ordered_file, util.maxmincalculated_file)
    myparams.generating_Test_Graph()
    topRank = Analyse.getTopRank(util.analysed_file + '.random.analised.txt')
    print 'Analising Files with TopRank', str(topRank)
    for OrderingFilePath in calc.getfilePathOrdered_separeted():
        analise = Analyse(myparams, OrderingFilePath, OrderingFilePath + '.analised.txt', topRank )