Put org in ignore as it has many FPs (#1200)

microsoft · Oct 31, 2023 · 22619f3 · 22619f3
1 parent 08c6c52
commit 22619f3
Show file tree

Hide file tree

Showing 5 changed files with 9 additions and 7 deletions.
diff --git a/.pipelines/templates/build-image-redactor.yml b/.pipelines/templates/build-image-redactor.yml
@@ -27,6 +27,7 @@ steps:
               set -eux  # fail on error
               pipenv install --deploy --dev
               pipenv run python -m spacy download en_core_web_lg
+              pipenv run pip install -e ../presidio-analyzer/. # Use the existing analyzer and not the one in PyPI
     - template: ./build-python.yml
       parameters:
           SERVICE: 'Image-Redactor'

diff --git a/presidio-analyzer/conf/default.yaml b/presidio-analyzer/conf/default.yaml
@@ -20,9 +20,9 @@ ner_model_configuration:
 
   low_confidence_score_multiplier: 0.4
   low_score_entity_names:
-  - ORG
-  - ORGANIZATION
+  -
   labels_to_ignore:
+  - ORGANIZATION # Has many false positives
   - CARDINAL
   - EVENT
   - LANGUAGE

diff --git a/presidio-analyzer/conf/spacy.yaml b/presidio-analyzer/conf/spacy.yaml
@@ -20,9 +20,10 @@ ner_model_configuration:
 
   low_confidence_score_multiplier: 0.4
   low_score_entity_names:
-  - ORG
-  - ORGANIZATION
+
   labels_to_ignore:
+  - ORG
+  - ORGANIZATION # has many false positives
   - CARDINAL
   - EVENT
   - LANGUAGE

diff --git a/presidio-analyzer/presidio_analyzer/nlp_engine/ner_model_configuration.py b/presidio-analyzer/presidio_analyzer/nlp_engine/ner_model_configuration.py
@@ -26,8 +26,8 @@
     HOSPITAL="ORGANIZATION",
 )
 
-LOW_SCORE_ENTITY_NAMES = {"ORG", "ORGANIZATION"}
-LABELS_TO_IGNORE = {"O"}
+LOW_SCORE_ENTITY_NAMES = {}
+LABELS_TO_IGNORE = {"O", "ORG", "ORGANIZATION"}
 
 
 @dataclass

diff --git a/presidio-image-redactor/tests/integration/test_image_analyzer_engine_integration.py b/presidio-image-redactor/tests/integration/test_image_analyzer_engine_integration.py
@@ -5,7 +5,7 @@
 def test_given_image_then_text_entities_are_recognized_correctly(image_analyzer_engine):
     # Image with PII entities
     image = get_resource_image("ocr_test.png")
-    analyzer_results = image_analyzer_engine.analyze(image)
+    analyzer_results = image_analyzer_engine.analyze(image, score_threshold=0.4)
     assert len(analyzer_results) == 7
     results = __get_expected_ocr_test_image_analysis_results()
     for i in range(7):