面向持续知识图谱嵌入中候选集干扰的匹配超额超越者正则化
文章背景与核心概要
持续知识图谱嵌入（Continual Knowledge Graph Embedding）旨在随着图谱随时间的增长，不断更新实体和关系的表示。尽管现有研究主要集中在缓解灾难性遗忘上，但本文指出了一个此前被忽视的关键挑战：候选集干扰（candidate-set interference）。当新实体被引入知识图谱时，它们会扩展所有兼容查询的候选实体空间。因此，即便历史正确答案的绝对得分及其在旧实体之间的相对顺序保持完全不变，其排名也可能会下降。
为了解决这一问题，作者提出了匹配超额超越者正则化（Matched Excess-Outranker Regularization, MEOR），这是一种新颖的主体级（host-level）优化目标。MEOR 通过单边惩罚机制，将平滑的答案相关新来者压力与分数盲区、结构匹配的旧参考进行对比，该惩罚仅在当新来者的竞争超过参考时触发。这确保了模型在继续学习合法新实体的有效表示的同时，不会降低历史排名。在多项基准测试和消融实验中，MEOR 展现出了卓越的性能提升和广泛的适用性。
📋 摘要与概要 (Summary)
持续知识图谱嵌入在图谱增长的过程中更新实体和关系表示。现有方法主要解决灾难性遗忘问题，但实体的引入同时也会改变每个兼容查询的候选空间。因此，即使历史答案的得分及其在旧实体中的排序得以保留，它也可能会失去排名。
Continual knowledge graph embedding updates entity and relation representations as a graph grows. Existing methods primarily address catastrophic forgetting, but entity admission also changes the candidate universe of every compatible query. A historical answer can therefore lose rank even when its score and its ordering among old entities are preserved.
我们将这种效应形式化为候选集干扰，并引入了匹配超额超越者正则化（MEOR），这是一种主体级优化目标，它将平滑的答案相关新来者压力与分数盲区、结构匹配的旧参考进行对比。其单边惩罚仅在当新来者竞争超过匹配参考时起作用，从而保留了主体学习器对合法新实体的信号。
We formalize this effect as candidate-set interference and introduce Matched Excess-Outranker Regularization (MEOR), a host-level objective that compares smooth answer-relative newcomer pressure with score-blind, structurally matched old references. Its one-sided penalty acts only when newcomer competition exceeds the matched reference, preserving the host learner's signal for legitimate new entities.
在 ENTITY-ComplEx 上的八次配对运行中，与经验回放相比，MEOR 将历史当前空间平均倒数排名（MRR）提高了 0.0057，将候选集干扰减少了 0.0055，其单边 95% 置信区间下限分别为 0.0052 和 0.0051。它满足旧空间排名的保留标准以及新实体的获取，并在历史当前空间 MRR 上优于持续校准、匹配最大正则化器（MMR）和不匹配旧正则化器（UOR）。直接消融实验支持其参考构建和聚合的每个组件。在所有十个报告的 FBInc-S 和 FBInc-L 主体与骨干网设置中，添加 MEOR 也改善了历史排名，每个配对的 95% 置信区间均不包含零。这些结果确立了候选实体引入作为持续排名损失的独特来源，并表明在不替换底层嵌入架构或持续学习器的情况下，可以对其进行有效控制。
Across eight paired runs on ENTITY-ComplEx, MEOR improves historical current-universe mean reciprocal rank (MRR) by 0.0057 over replay and reduces candidate-set interference by 0.0055, with one-sided 95% lower bounds of 0.0052 and 0.0051, respectively. It satisfies the preservation criteria for old-universe ranking and newcomer acquisition and improves historical current-universe MRR over persistent calibration, matched maximum regularizer (MMR), and unmatched old regularizer (UOR). Direct ablations support each component of its reference construction and aggregation. Adding MEOR also improves historical ranking in all ten reported FBInc-S and FBInc-L host and backbone settings, with every paired 95% confidence interval excluding zero. These results establish candidate admission as a distinct source of continual rank loss and show that it can be controlled without replacing the underlying embedding architecture or continual learner.
📌 摘要原文参考 (Abstract)
Continual knowledge graph embedding updates entity and relation representations as a graph grows. Existing methods primarily address catastrophic forgetting, but entity admission also changes the candidate universe of every compatible query. A historical answer can therefore lose rank even when its score and its ordering among old entities are preserved.
We formalize this effect as candidate-set interference and introduce Matched Excess-Outranker Regularization (MEOR), a host-level objective that compares smooth answer-relative newcomer pressure with score-blind, structurally matched old references. Its one-sided penalty acts only when newcomer competition exceeds the matched reference, preserving the host learner's signal for legitimate new entities.
Across eight paired runs on ENTITY-ComplEx, MEOR improves historical current-universe mean reciprocal rank (MRR) by 0.0057 over replay and reduces candidate-set interference by 0.0055, with one-sided 95% lower bounds of 0.0052 and 0.0051, respectively. It satisfies the preservation criteria for old-universe ranking and newcomer acquisition and improves historical current-universe MRR over persistent calibration, matched maximum regularizer (MMR), and unmatched old regularizer (UOR). Direct ablations support each component of its reference construction and aggregation. Adding MEOR also improves historical ranking in all ten reported FBInc-S and FBInc-L host and backbone settings, with every paired 95% confidence interval excluding zero. These results establish candidate admission as a distinct source of continual rank loss and show that it can be controlled without replacing the underlying embedding architecture or continual learner.