Replace all transcription scores with confidences

arkindex_worker/worker/transcription.py

@@ -18,7 +18,11 @@ class TextOrientation(Enum):
 
 class TranscriptionMixin(object):
     def create_transcription(
-        self, element, text, score, orientation=TextOrientation.HorizontalLeftToRight
+        self,
+        element,
+        text,
+        confidence,
+        orientation=TextOrientation.HorizontalLeftToRight,
     ):
         """
         Create a transcription on the given element through the API.
@@ -33,8 +37,8 @@ class TranscriptionMixin(object):
             orientation, TextOrientation
         ), "orientation shouldn't be null and should be of type TextOrientation"
         assert (
-            isinstance(score, float) and 0 <= score <= 1
-        ), "score shouldn't be null and should be a float in [0..1] range"
+            isinstance(confidence, float) and 0 <= confidence <= 1
+        ), "confidence shouldn't be null and should be a float in [0..1] range"
 
         if self.is_read_only:
             logger.warning(
@@ -48,7 +52,7 @@ class TranscriptionMixin(object):
             body={
                 "text": text,
                 "worker_version": self.worker_version_id,
-                "score": score,
+                "confidence": confidence,
                 "orientation": orientation.value,
             },
         )
@@ -99,10 +103,12 @@ class TranscriptionMixin(object):
                 text, str
             ), f"Transcription at index {index} in transcriptions: text shouldn't be null and should be of type str"
 
-            score = transcription.get("score")
+            confidence = transcription.get("confidence")
             assert (
-                score is not None and isinstance(score, float) and 0 <= score <= 1
-            ), f"Transcription at index {index} in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+                confidence is not None
+                and isinstance(confidence, float)
+                and 0 <= confidence <= 1
+            ), f"Transcription at index {index} in transcriptions: confidence shouldn't be null and should be a float in [0..1] range"
 
             orientation = transcription.get(
                 "orientation", TextOrientation.HorizontalLeftToRight
@@ -169,10 +175,12 @@ class TranscriptionMixin(object):
                 text, str
             ), f"Transcription at index {index} in transcriptions: text shouldn't be null and should be of type str"
 
-            score = transcription.get("score")
+            confidence = transcription.get("confidence")
             assert (
-                score is not None and isinstance(score, float) and 0 <= score <= 1
-            ), f"Transcription at index {index} in transcriptions: score shouldn't be null and should be a float in [0..1] range"
+                confidence is not None
+                and isinstance(confidence, float)
+                and 0 <= confidence <= 1
+            ), f"Transcription at index {index} in transcriptions: confidence shouldn't be null and should be a float in [0..1] range"
 
             orientation = transcription.get(
                 "orientation", TextOrientation.HorizontalLeftToRight
@@ -255,7 +263,7 @@ class TranscriptionMixin(object):
                         "id": annotation["id"],
                         "element_id": annotation["element_id"],
                         "text": transcription["text"],
-                        "confidence": transcription["score"],
+                        "confidence": transcription["confidence"],
                         "orientation": transcription.get(
                             "orientation", TextOrientation.HorizontalLeftToRight
                         ).value,

tests/data/line_transcriptions_small.json

@@ -8,7 +8,7 @@
       "id": "008691ae-8133-48c4-88d5-d4cc9f65c06c",
       "type": "line",
       "text": "J . Caron &",
-      "score": 0.4781,
+      "confidence": 0.4781,
       "zone": null,
       "source": null,
       "worker_version_id": "3ca4a8e3-91d1-4b78-8d83-d8bbbf487996",