fix(event_similarity_processor_v2): Stricter duplicate distance checking

k0machi · soyacz · commit 4bb3bf0a310b · 2026-03-19T12:03:26.000+01:00
V1 was using 0.95, use [-0.05,0.05] range for V2 to enable similarly
strict comparisons

Fixes ARGUS-112
diff --git a/argusAI/event_similarity_processor_v2.py b/argusAI/event_similarity_processor_v2.py
@@ -139,7 +139,7 @@ def _mark_event_is_duplicate(self, run_id: str, ts: datetime, severity: str, emb
                 ]
                 if len(dupe_embeddings) == 0:
                     return False
-                dupe_embedding = next((row for row, distance in dupe_embeddings if -0.1 < distance < 0.1), None)
+                dupe_embedding = next((row for row, distance in dupe_embeddings if -0.05 < distance < 0.05), None)
                 if not dupe_embedding:
                     return False
                 q = f"SELECT event_id FROM {SCTEvent.__table_name__} WHERE run_id = ? AND severity = ? AND ts = ?"

Original file line number	Diff line number	Diff line change
`@@ -139,7 +139,7 @@ def _mark_event_is_duplicate(self, run_id: str, ts: datetime, severity: str, emb`
`139`	`139`	`]`
`140`	`140`	`if len(dupe_embeddings) == 0:`
`141`	`141`	`return False`
`142`		`- dupe_embedding = next((row for row, distance in dupe_embeddings if -0.1 < distance < 0.1), None)`
	`142`	`+ dupe_embedding = next((row for row, distance in dupe_embeddings if -0.05 < distance < 0.05), None)`
`143`	`143`	`if not dupe_embedding:`
`144`	`144`	`return False`
`145`	`145`	`q = f"SELECT event_id FROM {SCTEvent.__table_name__} WHERE run_id = ? AND severity = ? AND ts = ?"`