public static void main(String[] args) throws Exception {
// Logger.getLogger(DocumentFeaturesDiff.class).setLevel(Level.ALL);
GateUtils.initGateKeepLog();
GateUtils.registerCzsemPlugin();
ProcessingResource eval =
new PRSetup.SinglePRSetup(LearningEvaluator.class)
.putFeature("keyASName", ":-)")
// .putFeature("responseASName", "lemma_flex")
.putFeature("responseASName", "flex")
.putFeature("keyAnnotationsAreInDocumentFeatures", true)
.putFeatureList("annotationTypes", "Lookup")
.putFeatureList("featureNames", "meshID")
.createPR();
SerialAnalyserController controller =
(SerialAnalyserController)
Factory.createResource(SerialAnalyserController.class.getCanonicalName());
controller.add(eval);
Corpus corpus = Factory.newCorpus(null);
corpus.populate(
new File("C:\\Users\\dedek\\Desktop\\bmc\\experiment\\analyzed").toURI().toURL(),
// new File("C:\\Users\\dedek\\Desktop\\bmca_devel").toURI().toURL(),
null,
"utf8",
false);
System.err.println("populated");
controller.setCorpus(corpus);
controller.execute();
}
public static void performExperiment(
DatasetFactory ds_factory,
double ds_reduce_retio,
String[] eval_annot_types,
int repeatCount,
int numFolds,
String results_file_name,
TrainTest... engines)
throws URISyntaxException, IOException, ExecutionException, ResourceInstantiationException,
PersistenceException, JDOMException, BenchmarkReportInputFileFormatException {
for (int a = 0; a < repeatCount; a++) {
for (String annot_type : eval_annot_types) {
logger.info(String.format("Performing evaluation for annotation type: %s", annot_type));
LearningEvaluator.CentralResultsRepository.repository.clear();
TimeBenchmarkUtils.enableGateTimeBenchmark();
final DataSet dataset =
new DataSetReduce(ds_factory.createDataset(annot_type), ds_reduce_retio);
MachineLearningExperiment experiment = new MachineLearningExperiment(dataset, engines);
if (numFolds == 1) {
Corpus testCorpus =
((CzechLawDataSet) ds_factory.createDataset(annot_type)).getTestCorpus();
experiment.trainOnly();
experiment.testOnly(testCorpus);
logger.info("saving results, counting time statistics...");
MachineLearningExperimenter.saveResults(results_file_name);
} else {
// experiment.crossValidation(numFolds);
experiment.controlledCrossValidation(
10, new File("../intlib/train-10-fold-cross/").toURI().toURL(), null, false);
logger.info("saving results, counting time statistics...");
MachineLearningExperimenter.saveResults(results_file_name);
}
GateUtils.deleteAllPublicGateResources();
}
}
}
public Corpus getTestCorpus() throws PersistenceException, ResourceInstantiationException {
DataStore ds = GateUtils.openDataStore(dataStore);
Corpus corpus = GateUtils.loadCorpusFormDatastore(ds, "test___1354799449330___3473");
return corpus;
}
