refactor: enhance duplicate annotation detection criteria

Expand the criteria for identifying and removing duplicate annotations
in the workbook to include `predicate` and `predicate_id` in addition to
`element_xpath`, `object`, and `object_id`. This more comprehensive
approach ensures the removal of truly redundant annotations.

src/spinneret/workbook.py

@@ -225,15 +225,17 @@ def delete_duplicate_annotations(
     :param workbook: The annotation workbook
     :returns: The workbook with duplicate annotations removed
     :notes: The function removes duplicate annotations based on the
-        following columns: `element_xpath`, `object`, `object_id`, `date`. The
-        most recent annotation is preferred to allow improvements to other
-        fields set by the annotator.
-
+        following columns: `element_xpath`, `predicate`, `predicate_id`,
+        `object`, and `object_id`. The most recent annotation, based on `date`,
+        is preferred to allow improvements to other fields set by the
+        annotator.
     """
     wb = workbook.sort_values("date", ascending=False)
     wb = wb.drop_duplicates(
         subset=[
             "element_xpath",
+            "predicate",
+            "predicate_id",
             "object",
             "object_id",
         ],

tests/test_workbook.py

@@ -135,18 +135,14 @@ def test_delete_duplicate_annotations():
         elements=["dataset"],
     )
     # Row 1
-    wb.loc[0, "predicate"] = "predicate_1"
-    wb.loc[0, "predicate_id"] = "predicate_id_1"
-    wb.loc[0, "object"] = "object_1"
-    wb.loc[0, "object_id"] = "object_id_1"
+    wb.loc[0, "predicate"] = "predicate"
+    wb.loc[0, "predicate_id"] = "predicate_id"
+    wb.loc[0, "object"] = "object"
+    wb.loc[0, "object_id"] = "object_id"
     wb.loc[0, "date"] = pd.Timestamp.now()
     # Row 2 is a duplicate annotation of row 1
     row = wb.iloc[0].copy()
     wb.loc[len(wb)] = row
-    wb.loc[1, "predicate"] = "predicate_2"
-    wb.loc[1, "predicate_id"] = "predicate_id_2"
-    wb.loc[1, "object"] = "object_1"
-    wb.loc[1, "object_id"] = "object_id_1"
     sleep(1)  # pause for 1 second to ensure the datetime is different
     wb.loc[1, "date"] = pd.Timestamp.now()