advoided misclassifcation Scigen-Physgen

src/fr/imag/forge/scidetect/Checker/Classifier.java

@@ -28,30 +28,33 @@ import fr.imag.forge.scidetect.Checker.Utils.DistancesSet;
 import fr.imag.forge.scidetect.Checker.Utils.ThresholdsSet;
 
 /**
- * Classifier is tagging input files has being of a certain class. Example of classes are SCIgen, Mathgen,...
- * The decision is made according to the distance between the tested file and its nearest neighbor 
- * in each class. Thresholds for assigning are read in file specified in the configuration file 
+ * Classifier is tagging input files has being of a certain class. Example of
+ * classes are SCIgen, Mathgen,... The decision is made according to the
+ * distance between the tested file and its nearest neighbor in each class.
+ * Thresholds for assigning are read in file specified in the configuration file
  * (default config.txt).
+ *
  * @author Nguyen Minh Tien - minh-tien.nguyen@imag.frs
  */
 public class Classifier {
 
-	private ThresholdsSet SetOfThresholds;
-	
-	/**
-	 * @throws Exception 
-	 * 
-	 */
-	public Classifier() {
-		this.SetOfThresholds = new ThresholdsSet();
-		this.SetOfThresholds.Init();
-	}
-	
-	
+    private ThresholdsSet SetOfThresholds;
+
+    /**
+     * @throws Exception
+     *
+     */
+    public Classifier() {
+        this.SetOfThresholds = new ThresholdsSet();
+        this.SetOfThresholds.Init();
+    }
+
     /**
-     * Classify is classifying each document given the matrix of distances (distant).
-     * For each entry it gives the class (or more) to which the text can be assigned 
-     * @param distant is a matrix of distances 
+     * Classify is classifying each document given the matrix of distances
+     * (distant). For each entry it gives the class (or more) to which the text
+     * can be assigned
+     *
+     * @param distant is a matrix of distances
      * @return the assigned class
      * @throws IOException
      */
@@ -59,7 +62,7 @@ public class Classifier {
 
         String result = "";
         String conclusion = "";
-        
+
         for (String key : distant.keySet()) {
             //for each file in the test
             result = find_NN(distant.get(key));
@@ -82,9 +85,12 @@ public class Classifier {
         return conclusion;
 
     }
+
     /**
-     * Check if the distance is lower, between of upper the two threshold. 
-     * @param result a string composed having for each classes the value of its NN 
+     * Check if the distance is lower, between of upper the two threshold.
+     *
+     * @param result a string composed having for each classes the value of its
+     * NN
      * @return
      */
     private String checkdistant(String result) {
@@ -96,6 +102,7 @@ public class Classifier {
             //System.out.println(eachtype[i]);
             //get threshold for the corresponding type
             Double[] threshold = new Double[2];
+
             if (SetOfThresholds.containsKey(eachNN[0])) {
                 threshold = SetOfThresholds.get(eachNN[0]);
             } else {
@@ -114,7 +121,10 @@ public class Classifier {
         }
         return conclution;
     }
+
     /**
+     * In case it is not a generation, find the closest text to it in general
+     *
      * @param result
      * @return
      */
@@ -132,10 +142,12 @@ public class Classifier {
         }
         return conclu;
     }
+
     /**
-     * 
-     * @param indexpath
-     * @return
+     * Get the type of the generation base on the path to it
+     *
+     * @param indexpath path to the NN
+     * @return type of generation (parent folder)
      */
     private String gettype(String indexpath) {
         File indexfile = new File(indexpath);
@@ -146,6 +158,8 @@ public class Classifier {
     }
 
     /**
+     * Get the nearest neighbor in each type of generation
+     *
      * @param distantto
      * @return
      */
@@ -166,6 +180,14 @@ public class Classifier {
             }
 
         }
+//remove either physgen or scigen to advoid misslassification
+        if (distotype.containsKey("Scigen") && distotype.containsKey("Physgen")) {
+            if (distotype.get("Scigen") < distotype.get("Physgen")) {
+                distotype.remove("Physgen");
+            } else {
+                distotype.remove("Scigen");
+            }
+        }
         // it returns the path to the NN
         String result = "";
         for (String key : distotype.keySet()) {