printPrecisions = False printDocuments = True numpy.set_printoptions(suppress=True, precision=3, linewidth=100) dataset = ArnetMinerDataset(runLSI=ranLSI) #dataset.fields = ["Intelligent Agents"] if printDocuments: print("Reading article data") authorList, documentList, citationList = dataset.readAuthorsAndDocuments(useAbstract=False) print("Done") ns = numpy.arange(5, 55, 5) bestaverageTestPrecisions = numpy.zeros(len(dataset.fields)) computeInfluence = True graphRanker = GraphRanker(k=100, numRuns=100, computeInfluence=computeInfluence, p=0.05, inputRanking=[1, 2]) methodNames = graphRanker.getNames() methodNames.append("MC2") numMethods = len(methodNames) averageTrainPrecisions = numpy.zeros((len(dataset.fields), len(ns), numMethods)) averageTestPrecisions = numpy.zeros((len(dataset.fields), len(ns), numMethods)) coverages = numpy.load(dataset.coverageFilename) print("==== Coverages ====") print(coverages) for s, field in enumerate(dataset.fields): if ranLSI: outputFilename = dataset.getOutputFieldDir(field) + "outputListsLSI.npz" documentFilename = dataset.getOutputFieldDir(field) + "relevantDocsLSI.npy"
testExpertMatchesInds = authorIndexer.translate(testExpertMatches) relevantAuthorInds1 = authorIndexer.translate(relAuthorsDocSimilarity) relevantAuthorInds2 = authorIndexer.translate(relAuthorsDocCitations) relevantAuthorsInds = authorIndexer.translate(relevantAuthors) expertAuthorsInds = authorIndexer1.translate(expertAuthors)#Get Ids our BM25 List assert (numpy.array(relevantAuthorInds1) < len(relevantAuthorsInds)).all() assert (numpy.array(relevantAuthorInds2) < len(relevantAuthorsInds)).all() if len(testExpertMatches) != 0: fich.write(field) fich.write("\n") #First compute graph properties computeInfluence = False #graphRanker = GraphRanker(k=100, numRuns=100, computeInfluence=computeInfluence, p=0.05, inputRanking=[relevantAuthorInds1, relevantAuthorInds2]) graphRanker = GraphRanker(k=100, numRuns=100, computeInfluence=computeInfluence, p=0.05, inputRanking=None) outputLists = graphRanker.vertexRankings(graph, relevantAuthorsInds) for line in outputLists: ListeAuthorsFinale = authorIndexer.reverseTranslate(line) #fich.write(' '.join(map(str, ListeAuthorsFinale))) fich.write(str(ListeAuthorsFinale)+"\n") outputListsE = [] #outputListsE.append(expertAuthorsInds)#Add BM25 Listto the outputList in order to aggregate scores later for line in outputListsE: ListeAuthorsFinale = authorIndexer1.reverseTranslate(line) #fich.write(' '.join(map(str, ListeAuthorsFinale))) fich.write(str(ListeAuthorsFinale)+"\n") #Save relevant authors #numpy.save(dataset.dataDir + "relevantAuthorsReputation" + field + ".txt", outputLists) fich.write("-----------------------------------------------")