Spaces:
Sleeping
Sleeping
Lenient thres
Browse files- search/search.py +13 -2
search/search.py
CHANGED
|
@@ -307,13 +307,24 @@ def search_comprehensive(query: str, num_results: int = 15, target_language: str
|
|
| 307 |
|
| 308 |
# Use reranker to improve overall quality and relevance
|
| 309 |
if all_results:
|
| 310 |
-
reranked_results = reranker.rerank_results(boosted_query, all_results, min_score=0.
|
| 311 |
logger.info(f"Reranked {len(all_results)} total results to {len(reranked_results)} high-quality results")
|
| 312 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 313 |
|
| 314 |
# Limit final results to requested count
|
| 315 |
all_results = all_results[:num_results]
|
| 316 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 317 |
# Create URL mapping
|
| 318 |
url_mapping = {}
|
| 319 |
for i, result in enumerate(all_results, 1):
|
|
|
|
| 307 |
|
| 308 |
# Use reranker to improve overall quality and relevance
|
| 309 |
if all_results:
|
| 310 |
+
reranked_results = reranker.rerank_results(boosted_query, all_results, min_score=0.1) # Much more lenient
|
| 311 |
logger.info(f"Reranked {len(all_results)} total results to {len(reranked_results)} high-quality results")
|
| 312 |
+
|
| 313 |
+
# If reranking filtered out too many results, use original results
|
| 314 |
+
if len(reranked_results) < max(1, len(all_results) // 4): # If less than 25% remain
|
| 315 |
+
logger.warning(f"Reranking too strict ({len(reranked_results)}/{len(all_results)}), using original results")
|
| 316 |
+
all_results = all_results[:num_results] # Just take top N original results
|
| 317 |
+
else:
|
| 318 |
+
all_results = reranked_results
|
| 319 |
|
| 320 |
# Limit final results to requested count
|
| 321 |
all_results = all_results[:num_results]
|
| 322 |
|
| 323 |
+
# Final safety check - ensure we have at least some results
|
| 324 |
+
if not all_results and text_results:
|
| 325 |
+
logger.warning("No results after processing, using original text results as fallback")
|
| 326 |
+
all_results = text_results[:num_results]
|
| 327 |
+
|
| 328 |
# Create URL mapping
|
| 329 |
url_mapping = {}
|
| 330 |
for i, result in enumerate(all_results, 1):
|