PR phrase labeling draft

git-svn-id: https://ws10smt.googlecode.com/svn/trunk@124 ec762483-ff6d-05da-a07a-a48fb63a330f

4 files changed, 13 insertions, 22 deletions

diff --git a/gi/posterior-regularisation/prjava/src/data/Corpus.java b/gi/posterior-regularisation/prjava/src/data/Corpus.java
index f0da0b33..425ede11 100644
--- a/gi/posterior-regularisation/prjava/src/data/Corpus.java
+++ b/gi/posterior-regularisation/prjava/src/data/Corpus.java
@@ -186,6 +186,9 @@ public class Corpus {
 				tagData.get(i)[j]=tagIdx(s[j]);

 			}

 		}

+		sent=null;

+		tag=null;

+		System.gc();

 	}

 	

 	public int [] getInt(int idx){

diff --git a/gi/posterior-regularisation/prjava/src/hmm/HMMObjective.java b/gi/posterior-regularisation/prjava/src/hmm/HMMObjective.java
index 551210c0..70b6c966 100644
--- a/gi/posterior-regularisation/prjava/src/hmm/HMMObjective.java
+++ b/gi/posterior-regularisation/prjava/src/hmm/HMMObjective.java
@@ -18,7 +18,7 @@ public class HMMObjective extends ProjectedObjective{
 	

 	private static final double GRAD_DIFF = 3;

 	public static double INIT_STEP_SIZE=10;

-	public static double VAL_DIFF=2000;

+	public static double VAL_DIFF=1000;

 	

 	private HMM hmm;

 	double[] newPoint  ;

@@ -33,7 +33,7 @@ public class HMMObjective extends ProjectedObjective{
 	private SimplexProjection projection;

 	

 	private int wordFreq[];

-	private static int MIN_FREQ=3;

+	private static int MIN_FREQ=10;

 	private int numWordsToProject=0;

 	

 	private int n_param;

@@ -88,6 +88,9 @@ public class HMMObjective extends ProjectedObjective{
 						if(projectionMap[word]==null){

 							projectionMap[word]=new TIntArrayList[n_states];

 						}

+			//			if(posteriorMap[sentNum][i]==null){

+			//				posteriorMap[sentNum][i]=new int[n_states];

+			//			}

 						

 						posteriorMap[sentNum][i][state]=n_param;

 						if(projectionMap[word][state]==null){

@@ -96,8 +99,8 @@ public class HMMObjective extends ProjectedObjective{
 						}

 						projectionMap[word][state].add(n_param);

 						n_param++;

-					}else{

-						

+					}

+					else{

 						posteriorMap[sentNum][i][state]=-1;

 					}

 				}

diff --git a/gi/posterior-regularisation/prjava/src/hmm/POS.java b/gi/posterior-regularisation/prjava/src/hmm/POS.java
index 722d38e2..2dcf271c 100644
--- a/gi/posterior-regularisation/prjava/src/hmm/POS.java
+++ b/gi/posterior-regularisation/prjava/src/hmm/POS.java
@@ -8,8 +8,8 @@ import data.Corpus;
 public class POS {

 

 	//public String trainFilename="../posdata/en_train.conll";

-	//public static String trainFilename="../posdata/small_train.txt";

-	public static String trainFilename="../posdata/en_test.conll";

+	public static String trainFilename="../posdata/small_train.txt";

+//	public static String trainFilename="../posdata/en_test.conll";

 //	public static String trainFilename="../posdata/trial1.txt";

 	

 	public static String testFilename="../posdata/en_test.conll";

@@ -72,21 +72,6 @@ public class POS {
 		}

 

 		hmm.writeModel(modelFilename);

-		

-		Corpus test=new Corpus(testFilename,c.vocab);

-		

-		PrintStream ps= io.FileUtil.openOutFile(predFilename);

-		

-		int [][]data=test.getAllData();

-		for(int i=0;i<data.length;i++){

-			int []tag=hmm.viterbi(data[i]);

-			String sent[]=test.get(i);

-			for(int j=0;j<data[i].length;j++){

-				ps.println(sent[j]+"\t"+tag[j]);

-			}

-			ps.println();

-		}

-		ps.close();

 	}

 	

 	

diff --git a/gi/posterior-regularisation/prjava/src/test/HMMModelStats.java b/gi/posterior-regularisation/prjava/src/test/HMMModelStats.java
index 26d7abec..dbf517fd 100644
--- a/gi/posterior-regularisation/prjava/src/test/HMMModelStats.java
+++ b/gi/posterior-regularisation/prjava/src/test/HMMModelStats.java
@@ -89,7 +89,7 @@ public class HMMModelStats {
 			}

 		}

 		

-		ps.println("max w t P(w_i|t)"+sum);

+		ps.println("max w t P(w_i|t): "+sum);

 		

 	}