fix: classify retraction nature for article flags [AI-assisted] (#1028)

Retraction Watch rows were previously cached as is_retracted=true regardless of RetractionNature. This caused corrections and expressions of concern to be reported as retracted articles.\n\nAdd explicit RetractionNature classification and regression tests so only true retraction-like records are flagged as retracted while preserving expected behavior for real retractions.

Co-authored-by: florath-ai-assistant[bot] <Andreas.Florath@telekom.de>

Author: coding-ai-assistant[bot]

src/aletheia_probe/updater/sources/retraction_watch.py

@@ -27,6 +27,14 @@
 ARTICLE_BATCH_SIZE = 1000
 PROGRESS_LOG_INTERVAL = 5000
 CACHE_EXPIRY_HOURS = 24 * 365
+NON_RETRACTION_NATURE_KEYWORDS = (
+    "correction",
+    "erratum",
+    "corrigendum",
+    "expression of concern",
+    "reinstatement",
+)
+RETRACTION_NATURE_KEYWORDS = ("retract", "withdraw")
 
 
 class RetractionWatchSource(DataSource):
@@ -549,11 +557,12 @@ def _collect_article_retractions(self, article_batch: list[dict[str, str]]) -> N
             retraction_nature = article.get("retraction_nature", "")
             reason = article.get("reason", "")
             retraction_doi = article.get("retraction_doi", "")
+            is_retracted = self._is_retracted_nature(retraction_nature)
 
             self.article_retractions.append(
                 {
                     "doi": doi.lower().strip(),
-                    "is_retracted": True,
+                    "is_retracted": is_retracted,
                     "retraction_type": retraction_nature or "Retraction",
                     "retraction_date": retraction_date_formatted,
                     "retraction_doi": retraction_doi if retraction_doi else None,
@@ -562,3 +571,15 @@ def _collect_article_retractions(self, article_batch: list[dict[str, str]]) -> N
                     "expires_at": expires_at.isoformat(),
                 }
             )
+
+    def _is_retracted_nature(self, retraction_nature: str) -> bool:
+        """Determine whether a RetractionNature value indicates true retraction."""
+        normalized = retraction_nature.lower().strip()
+        if not normalized:
+            # Keep historical behavior for blank types to avoid dropping valid retractions.
+            return True
+
+        if any(keyword in normalized for keyword in NON_RETRACTION_NATURE_KEYWORDS):
+            return False
+
+        return any(keyword in normalized for keyword in RETRACTION_NATURE_KEYWORDS)

tests/unit/updater/test_retraction_watch_source.py

@@ -92,3 +92,65 @@ async def test_fetch_data_missing_csv(self, source):
             with patch.object(source, "_clone_repository", return_value=repo_path):
                 result = await source.fetch_data()
                 assert result == []
+
+    @pytest.mark.parametrize(
+        ("retraction_nature", "expected"),
+        [
+            ("Retraction", True),
+            ("Partial Retraction", True),
+            ("Withdrawal", True),
+            ("Correction", False),
+            ("Erratum", False),
+            ("Corrigendum", False),
+            ("Expression of concern", False),
+            ("Reinstatement", False),
+            ("", True),
+        ],
+    )
+    def test_is_retracted_nature(self, source, retraction_nature, expected):
+        """Test RetractionNature classification for article-level retraction flag."""
+        assert source._is_retracted_nature(retraction_nature) is expected
+
+    def test_collect_article_retractions_marks_correction_not_retracted(self, source):
+        """Test correction records are cached as non-retracted article updates."""
+        source.article_retractions = []
+
+        source._collect_article_retractions(
+            [
+                {
+                    "doi": "10.1234/test-correction",
+                    "retraction_date_str": "01/15/2025 00:00",
+                    "retraction_nature": "Correction",
+                    "reason": "Error in figure",
+                    "retraction_doi": "10.1234/correction-notice",
+                }
+            ]
+        )
+
+        assert len(source.article_retractions) == 1
+        article = source.article_retractions[0]
+        assert article["doi"] == "10.1234/test-correction"
+        assert article["is_retracted"] is False
+        assert article["retraction_type"] == "Correction"
+
+    def test_collect_article_retractions_marks_retraction_as_retracted(self, source):
+        """Test retraction records remain cached as retracted articles."""
+        source.article_retractions = []
+
+        source._collect_article_retractions(
+            [
+                {
+                    "doi": "10.1234/test-retraction",
+                    "retraction_date_str": "01/15/2025 00:00",
+                    "retraction_nature": "Retraction",
+                    "reason": "Misconduct",
+                    "retraction_doi": "10.1234/retraction-notice",
+                }
+            ]
+        )
+
+        assert len(source.article_retractions) == 1
+        article = source.article_retractions[0]
+        assert article["doi"] == "10.1234/test-retraction"
+        assert article["is_retracted"] is True
+        assert article["retraction_type"] == "Retraction"