refactoring to improve the efficiency when metrics are computed

kermitt2 · Nov 17, 2017 · 99d3417 · 99d3417
1 parent 2107af1
commit 99d3417
Show file tree

Hide file tree

Showing 7 changed files with 350 additions and 344 deletions.
diff --git a/grobid-trainer/src/main/java/org/grobid/trainer/LabelStat.java b/grobid-trainer/src/main/java/org/grobid/trainer/LabelStat.java
@@ -6,36 +6,48 @@ public final class LabelStat {
     private int observed = 0; // this is true positives
     private int expected = 0; // total expected number of items with this label
 
+    private double accuracy = 0.0;
+    private int trueNegative;
+    private boolean hasChanged = false;
+
     public void incrementFalseNegative() {
         this.incrementFalseNegative(1);
+        hasChanged = true;
     }
 
     public void incrementFalsePositive() {
         this.incrementFalsePositive(1);
+        hasChanged = true;
     }
 
     public void incrementObserved() {
         this.incrementObserved(1);
+        hasChanged = true;
     }
 
     public void incrementExpected() {
         this.incrementExpected(1);
+        hasChanged = true;
     }
 
     public void incrementFalseNegative(int count) {
         this.falseNegative += count;
+        hasChanged = true;
     }
 
     public void incrementFalsePositive(int count) {
         this.falsePositive += count;
+        hasChanged = true;
     }
 
     public void incrementObserved(int count) {
         this.observed += count;
+        hasChanged = true;
     }
 
     public void incrementExpected(int count) {
         this.expected += count;
+        hasChanged = true;
     }
 
     public int getExpected() {
@@ -60,24 +72,36 @@ public int getAll() {
 
     public void setFalsePositive(int falsePositive) {
         this.falsePositive = falsePositive;
+        hasChanged = true;
     }
 
     public void setFalseNegative(int falseNegative) {
         this.falseNegative = falseNegative;
+        hasChanged = true;
     }
 
     public void setObserved(int observed) {
         this.observed = observed;
+        hasChanged = true;
     }
 
     public void setExpected(int expected) {
         this.expected = expected;
+        hasChanged = true;
     }
 
     public static LabelStat create() {
         return new LabelStat();
     }
 
+    public double getAccuracy() {
+        double accuracy = (double) (observed + trueNegative) / (observed + falsePositive + trueNegative + falseNegative);
+        if (accuracy < 0.0)
+            accuracy = 0.0;
+
+        return accuracy;
+    }
+
     public double getPrecision() {
         if (observed == 0.0) {
             return 0.0;
@@ -111,4 +135,14 @@ public String toString() {
             .append("; expected: ").append(expected);
         return builder.toString();
     }
+
+    public void setTrueNegative(int trueNegative) {
+        this.trueNegative = trueNegative;
+    }
+
+    public boolean hasChanged() {
+        Boolean oldValue = new Boolean(hasChanged);
+        hasChanged = false;
+        return oldValue;
+    }
 }
diff --git a/grobid-trainer/src/main/java/org/grobid/trainer/Stats.java b/grobid-trainer/src/main/java/org/grobid/trainer/Stats.java
diff --git a/grobid-trainer/src/main/java/org/grobid/trainer/evaluation/EndToEndEvaluation.java b/grobid-trainer/src/main/java/org/grobid/trainer/evaluation/EndToEndEvaluation.java
@@ -1,17 +1,11 @@
 package org.grobid.trainer.evaluation;
 
 import org.grobid.core.engines.config.GrobidAnalysisConfig;
-import org.grobid.core.engines.tagging.GenericTagger;
 import org.grobid.core.exceptions.*;
 import org.grobid.core.engines.Engine;
-import org.grobid.core.data.BiblioItem;
-import org.grobid.core.data.BibDataSet;
 import org.grobid.core.factory.GrobidFactory;
 import org.grobid.core.utilities.GrobidProperties;
 import org.grobid.core.utilities.UnicodeUtil;
-import org.grobid.trainer.Stats;
-import org.grobid.trainer.sax.NLMHeaderSaxHandler;
-import org.grobid.trainer.sax.FieldExtractSaxHandler;
 import org.grobid.core.utilities.TextUtilities;
 import org.grobid.trainer.evaluation.utilities.NamespaceContextMap;
 import org.grobid.trainer.evaluation.utilities.FieldSpecification;
@@ -26,7 +20,7 @@
 import javax.xml.xpath.XPathFactory;
 import javax.xml.parsers.*;
 import org.xml.sax.*;
-import org.xml.sax.helpers.*;
+
 import javax.xml.xpath.XPathConstants;
 
 import com.rockymadden.stringmetric.similarity.RatcliffObershelpMetric;