面向个人记忆检索的零推理前瞻性记忆项
原标题：Memory That Looks Forward: A Zero-Inference Prospective Term for Personal Memory Retrieval
AI 摘要
论文提出一种零推理成本的"前瞻性记忆检索"机制：将用户的承诺以带日期或触发条件的条目存入显式账本，当条目触发时，与之关联的记忆项获得显著性加成，并以乘法方式融入基于嵌入的检索，从而在查询时不运行任何模型。在仿照 TriggerBench 结构构建的合成任务集（48 段盲写对话、175 个任务）上，该机制将困难层的 recall@5 从 0.000 提升至默认权重下的 0.955、floor 变体下的 1.000，且在 53 个已解决承诺任务上零误加成。作者同时指出，只有 17–29% 的自然表述承诺-触发对会击败嵌入相似度，因此该机制只对少数真实案例起作用，但必须对其余案例无害。
正文节选
Memory That Looks Forward: A Zero-Inference Prospective Term for Personal Memory Retrieval Abstract Retrieval over a personal memory store is retrospective: it surfaces what resembles the query, and it is blind to what the user has committed to do. We describe a prospective term for memory retrieval that costs no inference at query time. Commitments are held in an explicit ledger as dated or trigger-conditioned entries; memory items linked to a firing entry receive a salience boost, blended mult