simon feedback

ls1intum · Dec 2, 2024 · b157de6 · b157de6
1 parent d363283
commit b157de6
Showing 1 changed file with 32 additions and 28 deletions.
diff --git a/src/main/java/de/tum/cit/aet/artemis/assessment/service/ResultService.java b/src/main/java/de/tum/cit/aet/artemis/assessment/service/ResultService.java
@@ -7,7 +7,7 @@
 import java.util.Collection;
 import java.util.Comparator;
 import java.util.HashMap;
-import java.util.Iterator;
+import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
 import java.util.Objects;
@@ -645,7 +645,7 @@ public FeedbackAnalysisResponseDTO getFeedbackDetailsOnPage(long exerciseId, Fee
             List<FeedbackDetailDTO> allFeedbackDetails = feedbackDetailPage.getContent();
 
             // Apply Levenshtein-based grouping and aggregation with a similarity threshold of 90%
-            List<FeedbackDetailDTO> aggregatedFeedbackDetails = aggregateUsingLevenshtein(allFeedbackDetails, 0.5);
+            List<FeedbackDetailDTO> aggregatedFeedbackDetails = aggregateUsingLevenshtein(allFeedbackDetails, 0.9);
 
             levenshteinMaxCount = aggregatedFeedbackDetails.stream().mapToLong(FeedbackDetailDTO::count).max().orElse(0);
             // Apply manual pagination
@@ -682,36 +682,40 @@ private Comparator<FeedbackDetailDTO> getComparatorForFeedbackDetails(FeedbackPa
     }
 
     private List<FeedbackDetailDTO> aggregateUsingLevenshtein(List<FeedbackDetailDTO> feedbackDetails, double similarityThreshold) {
-        // Group by testCaseName and taskName using nested maps
-        Map<String, Map<String, List<FeedbackDetailDTO>>> groupedByKeys = feedbackDetails.stream()
-                .collect(Collectors.groupingBy(FeedbackDetailDTO::testCaseName, Collectors.groupingBy(FeedbackDetailDTO::taskName)));
-
         List<FeedbackDetailDTO> aggregatedList = new ArrayList<>();
+        Set<FeedbackDetailDTO> processedDetails = new HashSet<>();
 
-        // Iterate through the grouped data
-        groupedByKeys.forEach((testCaseName, taskMap) -> {
-            taskMap.forEach((taskName, feedbackGroup) -> {
-                List<FeedbackDetailDTO> unprocessed = new ArrayList<>(feedbackGroup);
-                while (!unprocessed.isEmpty()) {
-                    FeedbackDetailDTO base = unprocessed.removeFirst();
-                    List<String> aggregatedTexts = new ArrayList<>(base.detailText());
-                    long totalCount = base.count();
-
-                    Iterator<FeedbackDetailDTO> iterator = unprocessed.iterator();
-                    while (iterator.hasNext()) {
-                        FeedbackDetailDTO compare = iterator.next();
-                        double similarity = NameSimilarity.levenshteinSimilarity(base.detailText().get(0), compare.detailText().get(0));
-                        if (similarity > similarityThreshold) {
-                            aggregatedTexts.add(compare.detailText().get(0));
-                            totalCount += compare.count();
-                            iterator.remove(); // Remove processed entry
-                        }
-                    }
+        for (FeedbackDetailDTO base : feedbackDetails) {
+            if (processedDetails.contains(base)) {
+                continue; // Skip already aggregated feedback
+            }
 
-                    aggregatedList.add(new FeedbackDetailDTO(totalCount, 0, aggregatedTexts, testCaseName, taskName, base.errorCategory()));
+            List<String> aggregatedTexts = new ArrayList<>();
+            aggregatedTexts.add(base.detailText().getFirst());
+            long totalCount = base.count();
+
+            for (FeedbackDetailDTO compare : feedbackDetails) {
+                if (processedDetails.contains(compare) || base.equals(compare)) {
+                    continue; // Skip already processed or the same feedback
                 }
-            });
-        });
+
+                // Ensure feedbacks have the same testCaseName and taskName
+                if (base.testCaseName().equals(compare.testCaseName()) && base.taskName().equals(compare.taskName())) {
+                    double similarity = NameSimilarity.levenshteinSimilarity(base.detailText().getFirst(), compare.detailText().getFirst());
+
+                    if (similarity > similarityThreshold) {
+                        // Merge the feedbacks
+                        aggregatedTexts.add(compare.detailText().getFirst());
+                        totalCount += compare.count();
+                        processedDetails.add(compare);
+                    }
+                }
+            }
+
+            // Add aggregated feedback entry
+            aggregatedList.add(new FeedbackDetailDTO(totalCount, 0, aggregatedTexts, base.testCaseName(), base.taskName(), base.errorCategory()));
+            processedDetails.add(base);
+        }
 
         return aggregatedList;
     }