@SuppressWarnings("unchecked")
public Set<String> processDoc(String str) throws Exception {
Set<String> toReturn = new HashSet<String>();
Corpus c = null;
Document aDoc = null;
try {
c = Factory.newCorpus("sample");
aDoc = Factory.newDocument(str);
c.add(aDoc);
controller.setCorpus(c);
controller.execute();
AnnotationSet aSet = aDoc.getAnnotations("StockSymbols");
for (Annotation annot : aSet) {
String symbol = (String) annot.getFeatures().get("sym");
toReturn.add(symbol);
}
} catch (Exception e) {
throw e;
} finally {
if (aDoc != null) {
Factory.deleteResource(aDoc);
}
if (c != null) {
Factory.deleteResource(c);
}
}
return toReturn;
}
/** Clear up the resources used after one test. */
private void clearOneTest() {
corpus.clear();
Factory.deleteResource(corpus);
Factory.deleteResource(learningApi);
controller.remove(learningApi);
controller.cleanup();
Factory.deleteResource(controller);
}
/**
* Loading the configurationg file and corpus for testing. And make settings as in the GATE Gui.
*/
void loadSettings(String configFileName, String corpusDirName, String inputasN, String outputasN)
throws GateException, IOException {
LogService.minVerbosityLevel = 0;
if (LogService.minVerbosityLevel > 0)
System.out.println("Learning Home : " + learningHome.getAbsolutePath());
FeatureMap parameters = Factory.newFeatureMap();
URL configFileURL = new File(configFileName).toURI().toURL();
parameters.put("configFileURL", configFileURL);
learningApi =
(LearningAPIMain) Factory.createResource("gate.learning.LearningAPIMain", parameters);
// Load the corpus
corpus = Factory.newCorpus("DataSet");
ExtensionFileFilter fileFilter = new ExtensionFileFilter();
fileFilter.addExtension("xml");
File[] xmlFiles = new File(corpusDirName).listFiles(fileFilter);
Arrays.sort(
xmlFiles,
new Comparator<File>() {
public int compare(File a, File b) {
return a.getName().compareTo(b.getName());
}
});
for (File f : xmlFiles) {
if (!f.isDirectory()) {
Document doc = Factory.newDocument(f.toURI().toURL(), "UTF-8");
doc.setName(f.getName());
corpus.add(doc);
}
}
// URL tempURL = new File(corpusDirName).toURI().toURL();
// corpus.populate(tempURL, fileFilter, "UTF-8", false);
// Set the inputAS
learningApi.setInputASName(inputasN);
learningApi.setOutputASName(outputasN);
controller =
(gate.creole.SerialAnalyserController)
Factory.createResource("gate.creole.SerialAnalyserController");
controller.setCorpus(corpus);
controller.add(learningApi);
}
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();
}
/**
* Test the chunk learning by using the Naive Bayes method and a small part of the OntoNews
* corpus.
*/
public void testNBChunkLearnng() throws IOException, GateException {
// Initialisation
System.out.print("Testing the Naive Bayes method on chunk learning...");
File chunklearningHome = new File(new File(learningHome, "test"), "chunklearning");
String configFileURL =
new File(chunklearningHome, "engines-naivebayesweka.xml").getAbsolutePath();
String corpusDirName = new File(chunklearningHome, "data-ontonews").getAbsolutePath();
// Remove the label list file, feature list file and chunk length files.
String wdResults =
new File(chunklearningHome, ConstantParameters.SUBDIRFORRESULTS).getAbsolutePath();
emptySavedFiles(wdResults);
String inputASN = "Key";
loadSettings(configFileURL, corpusDirName, inputASN, inputASN);
// Set the evaluation mode
RunMode runM = RunMode.EVALUATION;
learningApi.setLearningMode(runM);
controller.execute();
// Using the evaluation mode for testing
EvaluationBasedOnDocs evaluation = learningApi.getEvaluation();
// Compare the overall results with the correct numbers
/*assertEquals(evaluation.macroMeasuresOfResults.correct, 3);
assertEquals(evaluation.macroMeasuresOfResults.partialCor, 1);
assertEquals(evaluation.macroMeasuresOfResults.spurious, 19);
assertEquals(evaluation.macroMeasuresOfResults.missing, 68);*/
assertEquals(
"Wrong value for correct: ",
27,
(int) Math.floor(evaluation.macroMeasuresOfResults.correct));
assertEquals(
"Wrong value for partial: ",
3,
(int) Math.floor(evaluation.macroMeasuresOfResults.partialCor));
assertEquals(
"Wrong value for spurious: ",
26,
(int) Math.floor(evaluation.macroMeasuresOfResults.spurious));
assertEquals(
"Wrong value for missing: ",
42,
(int) Math.floor(evaluation.macroMeasuresOfResults.missing));
// Remove the resources
clearOneTest();
System.out.println("completed");
}