Merge pull request #36 from allenai/josh/claude/buoy-voltage-name-pattern

Add bare trailing-voltage pattern for buoy/gear name detection

Author: joshhvulcan

ais/src/atlantes/machine_annotation/data_annotate_utils.py

@@ -101,6 +101,11 @@ def get_ais_vessel_category(activity_descs: Optional[list[str]]) -> list[int]:
 #     reporting convention in gear names (~18% of GEAR records in VHS). The decimal
 #     voltage format (e.g. [7.7V]) adds ~1,355 incremental catches at 45% GEAR
 #     and only 1% FISHING.
+#   - Bare trailing voltage: "\d\s+\d+V\b" (e.g. "368000010   9V"). Whole-number
+#     volts ("9V", "12V") reported as a separate token after an ID/digit block,
+#     with no trailing decimal digit (so not caught by "\d+V\d+"). The leading
+#     "\d\s+" anchor requires preceding numeric content, which keeps the standalone
+#     "6V"/"7V" garbled-AIS case excluded (see NOT-included note below).
 #   - Double-dash separators: "\d+--\d+" (04001--2, 17002--41).
 #     Gear IDs frequently use double-dashes between numeric fields. Validated
 #     against VHS: 1,135 names not already caught by other patterns, of which
@@ -128,6 +133,7 @@ def get_ais_vessel_category(activity_descs: Optional[list[str]]) -> list[int]:
     r"\d+%",
     r"\d+V\d+",
     r"\d+\.\d+V",
+    r"\d\s+\d+V\b",
     r"\d+--\d+",
     r"Net fish",
     r"NetFish",

ais/tests/unit/inference/atlas_entity/test_unit_entity_postprocessor.py

@@ -716,6 +716,41 @@ def test_postprocess_numeric_gear_id_classified_as_buoy(
         assert output.entity_class == "buoy"
         assert output.entity_classification_details.postprocess_rule_applied is True
 
+    @pytest.mark.parametrize(
+        "entity_name",
+        [
+            "368000010         9V",
+            "368000002         9V",
+            "BUOY 992-536     12V",
+        ],
+    )
+    def test_postprocess_bare_trailing_voltage_classified_as_buoy(
+        self,
+        entity_name: str,
+        entity_postprocessor_class: AtlasEntityPostProcessor,
+    ) -> None:
+        """Test that names with a bare trailing voltage token (e.g. "9V", no decimal
+        digit) following a digit block are classified as buoys, even for USA MMSIs."""
+        input_data = EntityPostprocessorInput(
+            predicted_class=AtlasEntityLabelsTrainingWithUnknown.VESSEL,
+            entity_classification_details=EntityPostprocessorInputDetails(
+                model="test", confidence=0.9, outputs=[0.9, 0.1]
+            ),
+            metadata=EntityMetadata(
+                binned_ship_type=0,
+                ais_type=9999,
+                mmsi="368000010",
+                entity_name=entity_name,
+                track_length=800,
+                file_location=None,
+                trackId="A:368000010",
+                flag_code="USA",
+            ),
+        )
+        output = entity_postprocessor_class.postprocess(input_data)
+        assert output.entity_class == "buoy"
+        assert output.entity_classification_details.postprocess_rule_applied is True
+
     def test_postprocess_raises_error_for_known_binned_ship_type_and_buoy_name(
         self, entity_postprocessor_class: AtlasEntityPostProcessor
     ) -> None: