[论文] When Successful Knowledge Graph Edits Displace Correct Answers: Rank-L...
研究领域: ML 作者: Yi-Cheng Lai, Jerry Wang, Hsin-Ling Hsu, Li-Chu Chi, Ya-Wen Teng, Hen-Hsen Huang 发布时间: 2026-09-15 arXiv: 2609.12116
论文概要
研究领域: ML 作者: Yi-Cheng Lai, Jerry Wang, Hsin-Ling Hsu, Li-Chu Chi, Ya-Wen Teng, Hen-Hsen Huang 发布时间: 2026-09-15 arXiv: 2609.12116
中文摘要
编辑知识图谱嵌入(KGE)模型以提升期望答案,可能将正确答案从返回列表中排挤出去。仅基于复用被编辑参数的事实的局部性测试可能遗漏这种排序效应。我们在三个范围引入通用的排序位移审计:由被编辑参数支持的事实、目标查询的其他正确答案,以及具有相同关系的查询中的正确答案。我们还推导了使更新在精确保持选定分数的同时改进目标的维度与几何条件。在 FB15k-237 上使用 DistMult 与 ComplEx:直接提升总能将目标移入前十,但仅有 23.0–23.2% 的编辑是无损的;严格保持未造成可测损伤,但成功率仅为 1.3–1.4%;支持正则化的实体编辑取得最高的联合成功率 36.3–37.7%;截断排序的保持达到 32.8–34.7%,并将平均被排挤答案数从约 14 降至 1.2。跨维度、评分器、排序惯例以及一个学习型编辑器的实验表明,局部性同时取决于受保护范围与编辑机制。因此 KGE 编辑应同时报告修正成功率与排序位移的发生率和严重程度。
原文摘要
Editing a knowledge graph embedding (KGE) model to promote a desired answer can displace correct answers from the returned list. Locality tests based only on facts that reuse the edited parameter can miss this ranking effect. We introduce a common rank-displacement audit at three scopes: facts supported by the edited parameter, other correct answers to the target query, and correct answers across queries with the same relation. We also derive dimensional and geometric conditions for an update to improve the target while exactly preserving selected scores. On FB15k-237 with DistMult and ComplEx, direct promotion always moves the target into the top ten, but does so without damage in only 23.0--23.2\% of edits. Strict preservation causes no measured damage, yet succeeds in only 1.3--1.4\%. S...
*自动采集于 2026-09-15*
#论文 #arXiv #ML #小凯