SpeakerMem-R1：面向多方对话的说话人中心双轨记忆
原标题：Paper page - SpeakerMem-R1: Speaker-Centered Dual-Track Memory for Multi-Party Dialogue
AI 摘要
研究者提出 SpeakerMem-R1，一种以说话人为中心的双轨记忆框架，用于多方对话中的长期记忆。其双轨记忆分别存储带说话人标注的逐字消息和派生状态，并按人物级与群组级视图组织，查询时按实体、事件和时间融合两轨证据。团队用 SpeakerLevenshtein 奖励和说话人条件 GRPO 训练 Writer-R1（基于 Qwen2.5-3B），在 GroupMemBench、SocialMemBench、EverMemBench 上分别达到 47.9%、69.2%、61.9% 的二元准确率，并在 EverMemBench 公开榜单上取得 62.33% 的最佳结果。
正文节选
SpeakerMem-R1: Speaker-Centered Dual-Track Memory for Multi-Party Dialogue Abstract Long-term conversational memory in multi-party settings requires more than retrieving relevant content from long-term conversations: it must distinguish who said what, whom each statement concerns, how individuals perceive one another, what information is shared by the group, and how states change over time. Recent studies on multi-party dialogue benchmarks show that existing general-purpose LLM memory systems te