跳转至

文章背景与核心概要

现有的时序知识图谱补全方法通常假设实体、关系名称和时间戳在训练时是预定义且已知的，这极大地限制了模型只能应用于单一图谱和特定词表。为了解决这一局限性，本文提出了 FITTER，这是首个用于时序知识图谱链接预测的完全归纳式结构模型。它支持跨领域迁移，使得推理图能够包含来自不同领域的完全未见过的实体、关系和时间戳。

通过利用谓词与其他谓词的交互模式来表示它们，并采用时间平移不变的相对排序编码，FITTER 应用了消息传递机制，将局部和全局的时序上下文融合成与词表无关的嵌入表示。在六个不同的时序知识图谱基准测试上的评估表明，FITTER 在无需重新训练的情况下，持续优于现有的归纳式基线模型。

FITTER: Vocabulary-Agnostic Cross-Domain Inference on Temporal Knowledge Graphs

Summary

Existing temporal knowledge graph completion methods typically assume that entities, relation names, and timestamps are predefined and known during training, which limits models to a single graph and vocabulary. To address this limitation, FITTER introduces the first fully-inductive structural model for temporal knowledge graph link prediction. It enables cross-domain transfer where inference graphs can incorporate entirely unseen entities, relations, and timestamps from different domains.

By representing predicates through their interaction patterns and utilizing time-shift invariant relative ordering encodings, FITTER applies message-passing to fuse local and global temporal contexts into vocabulary-agnostic embeddings. Evaluated across six diverse temporal knowledge graph benchmarks, FITTER consistently outperforms inductive baselines without requiring retraining.

现有时序知识图谱补全方法通常假设实体、关系名称和时间戳在训练期间是预定义且已知的，这限制了模型只能处理单一图谱和词表。为了解决这一局限性，FITTER 推出了首个用于时序知识图谱链接预测的完全归纳式（fully-inductive）结构模型。它能够实现跨领域迁移，其中推理图可以包含来自不同领域的完全未见过的实体、关系和时间戳。

通过用谓词与其他谓词的交互模式来表示谓词，并利用时间平移不变的相对排序编码，FITTER 应用消息传递将局部和全局时序上下文融合成与词表无关的嵌入。在六个不同的时序知识图谱基准上进行评估后，FITTER 在无需重新训练的情况下始终优于归纳式基线。

Metadata & Publication Details

Metadata & Publication Details

元数据与出版详情

Authors

Authors

作者

  • Jiaxin Pan
  • Mojtaba Nayyeri
  • Osama Mohammed
  • Daniel Hernandez
  • Rongchuan Zhang
  • Cheng Cheng
  • Steffen Staab
  • Jiaxin Pan
  • Mojtaba Nayyeri
  • Osama Mohammed
  • Daniel Hernandez
  • Rongchuan Zhang
  • Cheng Cheng
  • Steffen Staab

Abstract

Abstract

摘要

Temporal knowledge graphs are central to many uses of the Semantic Web, but existing completion methods assume the entities, relation names, and timestamps to be reasoned about are already known at training time, restricting each model to a single graph and vocabulary. We propose FITTER, the first fully-inductive structural model for temporal knowledge graph link prediction that supports cross-domain transfer: the inference graph may contain entirely unseen entities, relation names, and timestamps drawn from a different domain. FITTER represents each predicate by its interaction patterns with others and time through encodings of relative rather than absolute ordering; message-passing fuses local and global temporal context to produce vocabulary-agnostic embeddings. We prove the temporal encoding is time-shift invariant and evaluate FITTER on cross-domain, cross-graph transfer over six temporal knowledge graph benchmarks of diverse domains, granularities, and time spans. FITTER consistently outperforms inductive baselines without retraining, indicating that vocabulary-agnostic structural learning is a viable foundation for inference over the heterogeneous knowledge graphs of the Semantic Web.

时序知识图谱是语义网许多应用的核心，但现有的补全方法假设在训练时要进行推理的实体、关系名称和时间戳是已知的，这限制了每个模型只能应用于单个图谱和词表。我们提出了 FITTER，这是首个用于时序知识图谱链接预测的完全归纳式结构模型，支持跨领域迁移：推理图可以包含来自不同领域的完全未见过的实体、关系名称和时间戳。FITTER 通过谓词与其他谓词的交互模式来表示每个谓词，并通过相对排序（而非绝对排序）的编码来表示时间；消息传递将局部和全局时序上下文融合成与词表无关的嵌入。我们证明了该时序编码具有时间平移不变性，并在跨领域、跨图谱迁移任务中，对跨越不同领域、粒度和时间跨度的六个时序知识图谱基准评估了 FITTER。FITTER 在无需重新训练的情况下始终优于归纳式基线，这表明与词表无关的结构学习是语义网异构知识图谱推理的可行基础。

Full-Text & External Resources

Full-Text & External Resources

全文与外部资源

Citations and Academic Tools

Citations and Academic Tools

引用与学术工具