Merge b662950 into 9d24bdc

bst-mug · Jun 24, 2019 · 47d8286 · 47d8286
2 parents 9d24bdc + b662950
commit 47d8286
Show file tree

Hide file tree

Showing 3 changed files with 7 additions and 13 deletions.
diff --git a/scripts/train_embeddings.sh b/scripts/train_embeddings.sh
@@ -2,4 +2,4 @@
 
 # Parameters follows https://github.com/ncbi-nlp/BioSentVec/blob/master/src/train_biowordvec.sh
 # `sentences.txt` should be first generated by `SentenceDumper.java`
-fasttext skipgram -input sentences.txt -output n2c2-fasttext -dim 200 -t 0.001 -minCount 0 -neg 10 -wordNgrams 6 -ws 30
+fasttext skipgram -input sentences.txt -output n2c2-fasttext -dim 200 -t 0.001 -minCount 0 -neg 10 -wordNgrams 6 -ws 20
diff --git a/src/main/java/at/medunigraz/imi/bst/n2c2/classifier/PerceptronClassifier.java b/src/main/java/at/medunigraz/imi/bst/n2c2/classifier/PerceptronClassifier.java
@@ -18,12 +18,13 @@ public class PerceptronClassifier extends CriterionBasedClassifier {
      * Location of precalculated vectors, extracted from the huge BioWordVec `.bin` file using `print_vectors.sh`.
      */
     private static final File PRETRAINED_VECTORS = new File(PerceptronClassifier.class.getClassLoader().getResource("BioWordVec-vectors.vec").getFile());
+    private static final File SELFTRAINED_VECTORS = new File(PerceptronClassifier.class.getClassLoader().getResource("self-trained-vectors.vec").getFile());
 
-    private final boolean preTrained;
+    private final File vectors;
 
     public PerceptronClassifier(Criterion c, boolean preTrained) {
         super(c);
-        this.preTrained = preTrained;
+        this.vectors = preTrained ? PRETRAINED_VECTORS : SELFTRAINED_VECTORS;
     }
 
     private String preprocess(String text) {
@@ -54,10 +55,6 @@ public void train(List<Patient> examples) {
             trainData.put(preprocess(p.getText()), p.getEligibility(criterion).name());
         }
 
-        if (preTrained) {
-            FastTextFacade.train(trainData, PRETRAINED_VECTORS);
-        } else {
-            FastTextFacade.train(trainData);
-        }
+        FastTextFacade.train(trainData, vectors);
     }
 }
diff --git a/src/main/java/at/medunigraz/imi/bst/n2c2/fasttext/FastTextFacade.java b/src/main/java/at/medunigraz/imi/bst/n2c2/fasttext/FastTextFacade.java
@@ -5,10 +5,7 @@
 import org.apache.logging.log4j.Logger;
 
 import java.io.*;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.List;
-import java.util.Map;
+import java.util.*;
 import java.util.concurrent.TimeUnit;
 import java.util.stream.Collectors;
 
@@ -133,7 +130,7 @@ private static List<String> getTrainCommand(File inputFile, File modelFile) {
         command.add("-epoch");
         command.add(String.valueOf(EPOCHS));
         command.add("-lr");
-        command.add(String.format("%.2f", LEARNING_RATE));
+        command.add(String.format(Locale.ROOT, "%.2f", LEARNING_RATE));
         command.add("-dim");
         command.add(String.valueOf(DIMENSIONS));
         return command;