[changes] Specified element quantity

This commit is contained in:
lanceyq
2026-03-27 12:03:48 +08:00
parent cc58c7333c
commit 8ba0a74473

View File

@@ -295,6 +295,7 @@ class ExtractionOrchestrator:
statement_entity_edges, statement_entity_edges,
entity_entity_edges, entity_entity_edges,
dialog_data_list, dialog_data_list,
dedup_details,
) = await self._run_dedup_and_write_summary( ) = await self._run_dedup_and_write_summary(
dialogue_nodes, dialogue_nodes,
chunk_nodes, chunk_nodes,
@@ -306,6 +307,11 @@ class ExtractionOrchestrator:
dialog_data_list, dialog_data_list,
) )
# 步骤 7: 同步用户别名到数据库表(仅正式模式)
if not is_pilot_run:
logger.info("步骤 7: 同步用户别名到 end_user 和 end_user_info 表")
await self._update_end_user_other_name(entity_nodes, dialog_data_list)
logger.info(f"知识提取流水线运行完成({mode_str}") logger.info(f"知识提取流水线运行完成({mode_str}")
return ( return (
dialogue_nodes, dialogue_nodes,
@@ -1492,6 +1498,7 @@ class ExtractionOrchestrator:
list[StatementChunkEdge], list[StatementChunkEdge],
list[StatementEntityEdge], list[StatementEntityEdge],
list[EntityEntityEdge], list[EntityEntityEdge],
list[DialogData],
dict dict
]: ]:
""" """
@@ -1555,6 +1562,8 @@ class ExtractionOrchestrator:
statement_chunk_edges, statement_chunk_edges,
dedup_statement_entity_edges, dedup_statement_entity_edges,
dedup_entity_entity_edges, dedup_entity_entity_edges,
dialog_data_list,
dedup_details,
) )
final_entity_nodes = dedup_entity_nodes final_entity_nodes = dedup_entity_nodes
@@ -1562,7 +1571,7 @@ class ExtractionOrchestrator:
final_entity_entity_edges = dedup_entity_entity_edges final_entity_entity_edges = dedup_entity_entity_edges
else: else:
# 正式模式:执行完整的两阶段去重 # 正式模式:执行完整的两阶段去重
result_tuple = await dedup_layers_and_merge_and_return( dedup_result_tuple = await dedup_layers_and_merge_and_return(
dialogue_nodes, dialogue_nodes,
chunk_nodes, chunk_nodes,
statement_nodes, statement_nodes,
@@ -1578,19 +1587,31 @@ class ExtractionOrchestrator:
# 解包返回值 # 解包返回值
( (
_, dialogue_nodes,
_, chunk_nodes,
_, statement_nodes,
final_entity_nodes, final_entity_nodes,
_, statement_chunk_edges,
final_statement_entity_edges, final_statement_entity_edges,
final_entity_entity_edges, final_entity_entity_edges,
dedup_details, dedup_details,
) = result_tuple ) = dedup_result_tuple
# 保存去重消歧的详细记录到实例变量 # 保存去重消歧的详细记录到实例变量
self._save_dedup_details(dedup_details, entity_nodes, final_entity_nodes) self._save_dedup_details(dedup_details, entity_nodes, final_entity_nodes)
result_tuple = (
dialogue_nodes,
chunk_nodes,
statement_nodes,
final_entity_nodes,
statement_chunk_edges,
final_statement_entity_edges,
final_entity_entity_edges,
dialog_data_list,
dedup_details,
)
logger.info( logger.info(
f"去重后: {len(final_entity_nodes)} 个实体节点, " f"去重后: {len(final_entity_nodes)} 个实体节点, "
f"{len(final_statement_entity_edges)} 条陈述句-实体边, " f"{len(final_statement_entity_edges)} 条陈述句-实体边, "