|
@@ -322,23 +322,26 @@ class DatasetRetrieval:
|
|
|
for thread in threads:
|
|
|
thread.join()
|
|
|
|
|
|
- if reranking_enable:
|
|
|
- # do rerank for searched documents
|
|
|
- data_post_processor = DataPostProcessor(tenant_id, reranking_mode,
|
|
|
- reranking_model, weights, False)
|
|
|
+ with measure_time() as timer:
|
|
|
+ if reranking_enable:
|
|
|
+ # do rerank for searched documents
|
|
|
+ data_post_processor = DataPostProcessor(
|
|
|
+ tenant_id, reranking_mode,
|
|
|
+ reranking_model, weights, False
|
|
|
+ )
|
|
|
|
|
|
- with measure_time() as timer:
|
|
|
all_documents = data_post_processor.invoke(
|
|
|
query=query,
|
|
|
documents=all_documents,
|
|
|
score_threshold=score_threshold,
|
|
|
top_n=top_k
|
|
|
)
|
|
|
- else:
|
|
|
- if index_type == "economy":
|
|
|
- all_documents = self.calculate_keyword_score(query, all_documents, top_k)
|
|
|
- elif index_type == "high_quality":
|
|
|
- all_documents = self.calculate_vector_score(all_documents, top_k, score_threshold)
|
|
|
+ else:
|
|
|
+ if index_type == "economy":
|
|
|
+ all_documents = self.calculate_keyword_score(query, all_documents, top_k)
|
|
|
+ elif index_type == "high_quality":
|
|
|
+ all_documents = self.calculate_vector_score(all_documents, top_k, score_threshold)
|
|
|
+
|
|
|
self._on_query(query, dataset_ids, app_id, user_from, user_id)
|
|
|
|
|
|
if all_documents:
|