Ajout paramétrage seuil d'annotation

@@ -51,6 +51,7 @@ public class PreprocessingRunner  implements ApplicationRunner {
     private Boolean ratings;
 
     private float annotatePercent;
+    private int annotateThreshold;
 
     @Override
     public void run(ApplicationArguments applicationArguments) throws Exception {
@@ -58,7 +59,7 @@ public class PreprocessingRunner  implements ApplicationRunner {
         setFilenames();
         List<AssociationElement> associationElements = loadAssociationElements(new File(dataDir, completeFilename));
 //        associationElements = cleanupSmallCounts(associationElements, 1, 1);
-        List<AssociationElement> annotableElements = removeFirstSmallCounts(associationElements, 1, 1);
+        List<AssociationElement> annotableElements = removeFirstSmallCounts(associationElements);
         List<Integer> annotateIndexes = chooseAnnotated(associationElements, annotableElements, 1, 1);
         writeSampleAndAnnotated(new File(dataDir, sampleFilename), new File(dataDir, annontatedFilename), annotateIndexes, associationElements, annotableElements);
     }
@@ -116,7 +117,7 @@ public class PreprocessingRunner  implements ApplicationRunner {
         return associationElements;
     }
 
-    private List<AssociationElement> removeFirstSmallCounts(List<AssociationElement> associationElements, int userSize, int itemSize) {
+    private List<AssociationElement> removeFirstSmallCounts(List<AssociationElement> associationElements) {
 
         boolean removedUser;
         boolean removedItem;
@@ -132,7 +133,7 @@ public class PreprocessingRunner  implements ApplicationRunner {
             itemIdSet = associationElements.stream().map(element -> element.getItemId()).collect(Collectors.toSet());
             for (Long itemId : itemIdSet) {
                 userCount = associationElements.stream().filter(element -> element.getItemId() == itemId).count();
-                if (userCount <= userSize) {
+                if (userCount <= annotateThreshold) {
                     associationElements = associationElements.stream().filter(element -> element.getItemId() != itemId).collect(Collectors.toList());
                     if (!removedItem) {
                         removedItem = true;
@@ -149,7 +150,7 @@ public class PreprocessingRunner  implements ApplicationRunner {
             userIdSet = associationElements.stream().map(element -> element.getUserId()).collect(Collectors.toSet());
             for (Long userId : userIdSet) {
                 itemCount = associationElements.stream().filter(element -> element.getUserId() == userId).count();
-                if (itemCount <= itemSize) {
+                if (itemCount <= annotateThreshold) {
                     associationElements = associationElements.stream().filter(element -> element.getUserId() != userId).collect(Collectors.toList());
                     if (!removedUser) {
                         removedUser = true;
@@ -314,6 +315,11 @@ public class PreprocessingRunner  implements ApplicationRunner {
                 annotatePercent = 1.0f;
             }
 
+            if (properties.containsKey("annotateThreshold")) {
+                annotateThreshold = Integer.parseInt(properties.getProperty("annotateThreshold"));
+            } else {
+                annotateThreshold = 1;
+            }
         } catch (IOException e) {
             throw new PreprocessingException("Can't read parameters properties file " + dataDir + parametersFilename, e);
         }