Skip to content

Commit dcedc93

Browse files
committed
Fixed broken labeling
1 parent de15453 commit dcedc93

File tree

1 file changed

+11
-13
lines changed

1 file changed

+11
-13
lines changed

src/main/java/ml/anon/recognition/machinelearning/service/AnnotationService.java

Lines changed: 11 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -48,11 +48,11 @@ public class AnnotationService implements IAnnotationService {
4848
private final static String basePath = AnnotationService.class.getResource(File.separator + "GermaNER").getPath() + File.separator;
4949

5050

51-
private final static String pathToTokenizedFile = ResourceUtil.getPath("GermaNER"+File.separator+"temp-file-to-annotate.txt");
52-
private final static String pathToConfig = ResourceUtil.getPath("GermaNER"+File.separator+"config.properties");
53-
private final static String pathToOuputFile =ResourceUtil.getPath("GermaNER"+File.separator+ "taggedFile.txt");
51+
private final static String pathToTokenizedFile = ResourceUtil.getPath("GermaNER" + File.separator + "temp-file-to-annotate.txt");
52+
private final static String pathToConfig = ResourceUtil.getPath("GermaNER" + File.separator + "config.properties");
53+
private final static String pathToOuputFile = ResourceUtil.getPath("GermaNER" + File.separator + "taggedFile.txt");
5454
private final static String pathToModel = basePath + "model";
55-
public final static String pathToTrainingFile = ResourceUtil.getPath("GermaNER"+File.separator+ "trainingsFile.txt");
55+
public final static String pathToTrainingFile = ResourceUtil.getPath("GermaNER" + File.separator + "trainingsFile.txt");
5656
static Properties prop;
5757
static InputStream configFile = null;
5858
static File modelDirectory;
@@ -207,7 +207,7 @@ public static void initGermaNER() {
207207

208208
setModelDir();
209209

210-
File outputtmpFile = File.createTempFile("result",".tmp");
210+
File outputtmpFile = File.createTempFile("result", ".tmp");
211211
File outputFile = new File(pathToOuputFile);
212212

213213
long initNerModelB = System.currentTimeMillis();
@@ -249,15 +249,14 @@ public List<Anonymization> annotate(Document document) {
249249
}
250250

251251

252-
253252
File outputFile = new File(pathToOuputFile);
254253

255254
System.out.println("Start tagging");
256255

257256
PrintWriter out;
258257
ArrayList<Anonymization> anonymizations = null;
259258
try (InputStream inputStream = new FileInputStream(outputFile.getAbsolutePath())) {
260-
File outputtmpFile = File.createTempFile("result",".tmp");
259+
File outputtmpFile = File.createTempFile("result", ".tmp");
261260
out = new PrintWriter(pathToTokenizedFile);
262261
out.println(tokenizedFile);
263262
out.close();
@@ -319,18 +318,17 @@ private ArrayList<Anonymization> receiveAnonymizations(InputStream inputStream)
319318
original = original.trim();
320319
temp = counter;
321320
anonymization.original(original);
322-
try {
323-
anonymization.replacement(replacementGenerator.generateReplacement(original, label));
324-
} catch (Exception e) {
325-
e.printStackTrace();
326-
}
321+
322+
anonymization.replacement(replacementGenerator.generateReplacement(original, label));
323+
327324
anonymization.label(label);
328325

329326
anonymizations.add(anonymization.build());
330327
original = "";
331328

332329
}
333-
label = Label.valueOf(splitted[1].substring(2)); // Label - must exactly match!
330+
String substring = splitted[1].substring(2);
331+
label = Label.getOrDefault(substring, Label.UNKNOWN); // Label - must exactly match!
334332

335333
original += " " + splitted[0]; // 1. Teil des Tags
336334
log.debug(line);

0 commit comments

Comments
 (0)