bat-6 commited on
Commit
5b4a91e
·
1 Parent(s): 41bd215
src/similarity_model/similarity_engine.py CHANGED
@@ -287,22 +287,38 @@ def find_similar_projects(
287
 
288
  if len(existing_project) > 0:
289
 
290
- print("Using stored features")
291
 
292
- final_features = existing_project.iloc[0]["features"]
 
293
 
294
- else:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
295
 
296
- print("Extracting new features")
297
-
298
- final_features = merge_features(
299
- title=title,
300
- abstract=abstract,
301
- description=description,
302
- features=features,
303
- auto_extract=auto_extract_features
304
- )
305
 
 
 
 
 
 
 
 
306
 
307
 
308
  query_project = build_query_project(
@@ -384,6 +400,12 @@ def find_similar_projects(
384
  ascending=False
385
  ).head(top_k).reset_index(drop=True)
386
 
 
 
 
 
 
 
387
  # add debug info
388
  final_df["query_features_used"] = [
389
  final_features
 
287
 
288
  if len(existing_project) > 0:
289
 
290
+ logger.info("Exact title match found — returning originality = 0")
291
 
292
+ matched_row = existing_project.iloc[0]
293
+ stored_features = safe_feature_list(matched_row["features"])
294
 
295
+ return pd.DataFrame([{
296
+ "project_id": int(matched_row.name),
297
+ "project_title": matched_row["project_title"],
298
+ "semantic_score": 1.0,
299
+ "feature_score": 1.0,
300
+ "coverage": 1.0,
301
+ "hybrid_score": 1.0,
302
+ "originality_score": 0.0,
303
+ "confidence_score": 1.0,
304
+ "duplicate_risk": "Very High",
305
+ "shared_features_count": len(stored_features),
306
+ "matched_features": [{"feature_a": f, "feature_b": f, "score": 1.0} for f in stored_features],
307
+ "unique_query_features": [],
308
+ "unique_candidate_features": [],
309
+ "query_features_used": stored_features,
310
+ "query_clean_text": title
311
+ }])
312
 
313
+ logger.info("Extracting new features")
 
 
 
 
 
 
 
 
314
 
315
+ final_features = merge_features(
316
+ title=title,
317
+ abstract=abstract,
318
+ description=description,
319
+ features=features,
320
+ auto_extract=auto_extract_features
321
+ )
322
 
323
 
324
  query_project = build_query_project(
 
400
  ascending=False
401
  ).head(top_k).reset_index(drop=True)
402
 
403
+ # Force originality = 0 for any near-duplicate (same content, different title)
404
+ final_df.loc[
405
+ final_df["hybrid_score"] >= 0.90,
406
+ "originality_score"
407
+ ] = 0.0
408
+
409
  # add debug info
410
  final_df["query_features_used"] = [
411
  final_features