Update

Author: Henry Chen

src/main/java/edu/emory/clir/clearnlp/extraction/attribute/corpus/EntityTokenCorpusReconstructor.java

@@ -71,7 +71,7 @@ public EntityTokenCorpusReconstructor(CorpusType type, List<String> filePaths, S
 	private void init(){
 		switch (type) {
 			case RAW:	decoder = new NLPDecoder(TLanguage.ENGLISH);				break;
-			case TSV:	reader = new TSVReader(0, 1, 2, 3, 7, 4, 5, 6, -1, -1); 	break;
+			case TSV:	reader = new TSVReader(0, 1, 2, 3, 9, 4, 5, 6, 7, 8); 		break;
 		}
 	}
 

src/main/java/edu/emory/clir/clearnlp/extraction/attribute/corpus/MultiTheadedCorpusReconstructor.java

@@ -55,7 +55,7 @@ public void reconstruct(List<String> filePaths) {
 		AbstractCorpusRecontructor constructor_new;
 		ExecutorService executor = Executors.newFixedThreadPool(threadCount);
 		for(i = 0; i < l_size; i+=b_size){
-			if(i+b_size >= l_size){
+			if(i+b_size < l_size){
 				constructor_new = constructor.clone();
 				constructor_new.setFilePaths(filePaths.subList(i, i+b_size));
 				executor.submit(constructor_new);
@@ -65,7 +65,6 @@ public void reconstruct(List<String> filePaths) {
 				executor.submit(constructor);
 			}
 		}
-		
 	}
 
 	@Override

src/main/java/edu/emory/clir/clearnlp/extraction/attribute/ngram/AbstractNGram.java

@@ -20,22 +20,27 @@
 import org.junit.Test;
 
 import edu.emory.clir.clearnlp.util.DSUtils;
+import edu.emory.clir.clearnlp.util.FileUtils;
 
 /**
  * @author 	Yu-Hsin(Henry) Chen ({@code [email protected]})
  * @version	1.0
  * @since 	Sep 23, 2015
  */
 public class EntityTokenCorpusReconstructorTest{
-	public final Set<String> EXT = DSUtils.toHashSet(".txt");
-	public final String INPUT_DIR = "/Users/HenryChen/Documents/clearnlp-qa/corpus/NYT";
-	public final String OUPUT_DIR = "/Users/HenryChen/Documents/clearnlp-qa/corpus/NYT_OUT";
+	public final Set<String> EXT = DSUtils.toHashSet(".dep");
+	public final String INPUT_DIR = "/Users/HenryChen/Documents/clearnlp-qa/corpus/WSJ";
+	public final String OUPUT_DIR = "/Users/HenryChen/Documents/clearnlp-qa/corpus/WSJ_OUT";
 	public final Set<String> NERLabels = DSUtils.toHashSet("PERSON", "ORG", "LOC", "GPE");
 
 	@Test
 	public void testReconstructor(){
+		System.out.println(INPUT_DIR);
+		System.out.println(FileUtils.getBaseName(INPUT_DIR));
+		System.out.println(FileUtils.getFileList(INPUT_DIR, ".dep", false));
+		
 		EntityTokenCorpusReconstructor constructor 
-			= new EntityTokenCorpusReconstructor(CorpusType.RAW, INPUT_DIR, OUPUT_DIR, EXT, NERLabels);
+			= new EntityTokenCorpusReconstructor(CorpusType.TSV, INPUT_DIR, OUPUT_DIR, EXT, NERLabels);
 
 		constructor.reconstruct();
 	}