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();
}
/**
* @param args
* @throws Exception
*/
public static void main(String[] args) throws Exception {
/* Parse command line arguments */
Getopt g = new Getopt("gateExtractor", args, "i:r:te");
g.setOpterr(false);
String inputPath = "";
String outputPath = "";
boolean train = false;
boolean eval = false;
boolean run = false;
int c;
String arg;
while ((c = g.getopt()) != -1) {
switch (c) {
case 'i':
arg = g.getOptarg();
if (arg == null || arg.isEmpty()) {
usage("Please provide an input path");
}
inputPath = arg;
break;
case 'r':
run = true;
arg = g.getOptarg();
if (arg == null || arg.isEmpty()) {
usage("Please provide an output path");
}
outputPath = arg;
break;
case 't':
train = true;
break;
case 'e':
eval = true;
break;
case '?':
default:
usage(null);
}
}
if (args.length == 0 || (!run && !train && !eval)) {
usage("Nothing to do.");
}
if (inputPath == null || inputPath.isEmpty()) {
usage("Please provide an input path");
}
if (run && (outputPath == null || outputPath.isEmpty())) {
usage("Please provide an output directory!");
}
if (train && eval) {
usage("Only one mode allowed at a time");
}
if (train && run) {
usage("Only one mode allowed at a time");
}
if (eval && run) {
usage("Only one mode allowed at a time");
}
/* Initialize GATE */
String location =
new File(Main.class.getProtectionDomain().getCodeSource().getLocation().getPath())
.getParent();
String resourcesFolder = location + "/resources";
Gate.setGateHome(new File(resourcesFolder));
/* Create ml-config.xml with threads */
createConfig(resourcesFolder + File.separator);
Gate.init();
/* Load Corpus */
log.info("Loading Corpus ... ");
Corpus corpus = Factory.newCorpus("Training Corpus");
File directory = new File(inputPath);
URL url = directory.toURI().toURL();
corpus.populate(url, null, null, true);
log.info("Done loading Corpus!");
Pipeline pipeline = null;
/* Do Tagging */
pipeline = new Tagger();
pipeline.run(corpus, resourcesFolder);
/* Train */
if (train) {
pipeline = new Trainer();
pipeline.run(corpus, resourcesFolder);
}
/* Apply learned rules */
if (run) {
pipeline = new Extractor();
pipeline.run(corpus, resourcesFolder);
ExecutorService executorService = Executors.newFixedThreadPool(20);
for (int i = 0; i < corpus.size(); i++) {
executorService.execute(new OutputGenerator(outputPath, corpus.get(i)));
}
executorService.shutdown();
executorService.awaitTermination(Long.MAX_VALUE, TimeUnit.DAYS);
}
/* Evaluate results */
if (eval) {
pipeline = new Evaluator();
pipeline.run(corpus, resourcesFolder);
}
/* Clean up */
Factory.deleteResource(corpus);
outputFile_mlConfigThreads.delete();
}
