adai/MemoryBear
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[changes] Modify the index and complete the range

Browse Source
This commit is contained in:
lanceyq
2026-03-19 12:49:36 +08:00
parent c2fc4ab4ff
commit 2e50e30071
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
api/app/core/memory/storage_services/extraction_engine/data_preprocessing/data_pruning.py
View File

@@ -167,7 +167,7 @@ class SemanticPruner:
# 规则1：确认词 + 感谢词（如"好的谢谢"、"嗯谢谢"） # 规则1：确认词 + 感谢词（如"好的谢谢"、"嗯谢谢"）
for cp in _confirm_prefixes: for cp in _confirm_prefixes:
for ts in _thanks_suffixes: for ts in _thanks_suffixes:
if t == cp + ts or t == cp + "，" + ts or t == cp + "，" + ts: if t == cp + ts or t == cp + "，" + ts or t == cp + "," + ts:
return True return True
# 规则2：称呼前缀 + 问候（如"同学你好"、"老师好"） # 规则2：称呼前缀 + 问候（如"同学你好"、"老师好"）
@@ -760,7 +760,7 @@ class SemanticPruner:
# 相关对话：根据阶段决定处理力度 # 相关对话：根据阶段决定处理力度
if extraction.is_related: if extraction.is_related:
stats["related_count"] += 1 stats["related_count"] += 1
stats["related_indices"].append(d_idx) stats["related_indices"].append(d_idx + 1)
kept = self._apply_related_dialog_pruning( kept = self._apply_related_dialog_pruning(
msgs, extraction, f"对话 {d_idx+1}", pruning_mode msgs, extraction, f"对话 {d_idx+1}", pruning_mode
) )
@@ -778,7 +778,7 @@ class SemanticPruner:
continue continue
stats["unrelated_count"] += 1 stats["unrelated_count"] += 1
stats["unrelated_indices"].append(d_idx) stats["unrelated_indices"].append(d_idx + 1)
# 从 LLM 抽取结果中获取所有需要保留的 token # 从 LLM 抽取结果中获取所有需要保留的 token
preserve_tokens = self._build_preserve_tokens(extraction) preserve_tokens = self._build_preserve_tokens(extraction)