feat(provenance): ML stops surfacing the artist category

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-05-18 21:36:43 -04:00
parent e798302cfa
commit 592c665701
4 changed files with 34 additions and 4 deletions
-1
View File
@@ -23,7 +23,6 @@ from .embedder import MODEL_VERSION as SIGLIP_VERSION
ELIGIBLE_KINDS = {
TagKind.character,
TagKind.artist,
TagKind.fandom,
TagKind.general,
TagKind.series,
+3 -2
View File
@@ -48,12 +48,13 @@ class SuggestionService:
).scalar_one()
def _threshold_for(self, s: MLSettings, category: str) -> float:
# 'artist' intentionally absent (FC-2d-vii-c) — falls through to
# the 1.01 "never surfaces" default like any unsurfaced category.
return {
"artist": s.suggestion_threshold_artist,
"character": s.suggestion_threshold_character,
"copyright": s.suggestion_threshold_copyright,
"general": s.suggestion_threshold_general,
}.get(category, 1.01) # 1.01 => never surfaces (unsurfaced category)
}.get(category, 1.01)
async def for_image(self, image_id: int) -> SuggestionList:
img = await self.session.get(ImageRecord, image_id)
+4 -1
View File
@@ -34,7 +34,10 @@ STORE_FLOOR = float(os.environ.get("TAGGER_STORE_FLOOR", "0.05"))
# The categories FC-2b surfaces in the UI. Others (meta/rating/year) are
# still stored but the suggestion service filters them out.
SURFACED_CATEGORIES = {"artist", "character", "copyright", "general"}
# FC-2d-vii-c: 'artist' retired — artist identity is acquisition-derived
# (image_record.artist_id), never ML-inferred. Raw predictions are still
# stored at STORE_FLOOR but artist never surfaces.
SURFACED_CATEGORIES = {"character", "copyright", "general"}
@dataclass(frozen=True)
+27
View File
@@ -0,0 +1,27 @@
import pytest
pytestmark = pytest.mark.integration
def test_artist_not_surfaced():
from backend.app.services.ml.tagger import SURFACED_CATEGORIES
assert "artist" not in SURFACED_CATEGORIES
def test_artist_not_centroid_eligible():
from backend.app.models import TagKind
from backend.app.services.ml.centroids import ELIGIBLE_KINDS
assert TagKind.artist not in ELIGIBLE_KINDS
def test_threshold_for_artist_is_unsurfaced():
from backend.app.services.ml.suggestions import SuggestionService
class _S:
suggestion_threshold_character = 0.5
suggestion_threshold_copyright = 0.5
suggestion_threshold_general = 0.5
svc = SuggestionService.__new__(SuggestionService)
# 'artist' must fall through to the 1.01 "never surfaces" default
assert svc._threshold_for(_S(), "artist") == 1.01