文献类型：期刊文献

中文题名：带有差异化机制的多视角归纳式知识图谱补全框架

英文题名：Multi-view Framework for Inductive Knowledge Graph Completion with Differentiation Mechanism

作者：童翰文[1,2];钱羽希[4];刘井平[3];梁祖杰[4];肖仰华[1,2];韦峰[4];郝正鸿[4];韩冰[4]

机构：[1]复旦大学计算机科学技术学院,上海200438;[2]上海市数据科学重点实验室(复旦大学),上海200438;[3]华东理工大学信息科学与工程学院,上海200237;[4]蚂蚁集团,上海200010

年份：2025

卷号：36

期号：12

起止页码：5629

中文期刊名：软件学报

外文期刊名：Journal of Software

收录：;EI(收录号：20262721035789);北大核心:【北大核心2023】；

基金：国家自然科学基金青年科学基金(62306112);上海市青年科技英才扬帆计划(23YF1409400);上海市基础研究特区计划(22TQ1400100-20)。

语种：中文

中文关键词：归纳式知识图谱补全;多视角框架;差异化机制;知识图谱

外文关键词：inductive knowledge graph completion;multi-view framework;differentiation mechanism;knowledge graph

摘要：知识图谱补全模型需要具备归纳能力,才能够随着知识图谱的扩充泛化到新实体上.然而,现有的方法都只能通过聚合知识图谱中的邻居信息,从一个局部的视角来理解实体的语义,从而导致无法从不同的视角捕捉到实体之间的多种有价值的关联.在局部视角以外,通过非显式连接实体之间和远距离连接实体之间的交互,从而以全局视角和序列视角来进一步理解实体是至关重要的.更重要的是,强调通过多个不同视角聚合到的信息应当是互补的,而不是冗余的.因此,提出一个带有差异化机制的多视角知识图谱补全框架,用于归纳式知识图谱补全任务.它能够从多个不同视角学习到互补的、互不重叠的实体表示.具体来说,除了通过关系图卷积网络聚合邻居信息得到实体的局部表示外,设计一种基于注意力的差异化机制,用于从语义相关的实体和实体相关路径中聚合得到实体的全局和序列表示.最终,融合这些表示,并基于它们给三元组打分.实验结果证明,所提方法在归纳式的设定下超越了当前最先进的方法.此外,所提方法在直推式的知识图谱补全任务中也保持着有竞争力的表现.
Knowledge graph completion(KGC)models require inductive ability to generalize to new entities as the knowledge graph expands.However,current approaches understand entities only from a local perspective by aggregating neighboring information,failing to capture valuable interconnections between entities across different views. This study argues that global and sequential perspectives are essential for understanding entities beyond the local view by enabling interaction between disconnected and distant entity pairs. More importantly, it emphasizes that the aggregated information must be complementary across different views to avoid redundancy. Therefore, a multi-view framework with the differentiation mechanism is proposed for inductive KGC, aimed at learning complementary entity representations from various perspectives. Specifically, in addition to aggregating neighboring information to obtain the entity’s local representation through R-GCN, an attention-based differentiation mechanism is employed to aggregate complementary information from semantically related entities and entity-related paths, thus obtaining global and sequential representations of the entities. Finally, these representations are fused and used to score the triples. Experimental results demonstrate that the proposed framework consistently outperforms state-of-the-art approaches in the inductive setting. Moreover, it retains competitive performance in the transductive setting.