Generalize to Fully Unseen Graphs: Learn Transferable Hyper-Relation Structures for Inductive Link Prediction

Jing Yang, XiaowenJiang, Yuan Gao, Laurence Tianruo Yang, JieMing Yang

Abstract: Inductive link prediction aims to infer missing triples on unseen graphs, which contain unseen entities and relations during training. The performances of existing inductive inference methods were hindered by the limited generalization capability in fully unseen graphs, which is rooted in the neglect of the intrinsic graph structure. In this paper, we aim to enhance the model's generalization ability to unseen graphs and thus propose a novel Hyper-Relation aware multi-views model HyRel for learning the global transferable structure of graphs. Distinct from existing studies, we introduce a novel perspective focused on learning the inherent hyper-relation structure consisting of the relation positions and affinity. The hyper-relation structure is independent of specific entities, relations, or features, thus allowing for transferring the learned knowledge to any unseen graphs. We adopt a multi-view approach to model the hyper-relation structure. HyRel incorporates neighborhood learning on each view, capturing nuanced semantics of relative relation position. Meanwhile, dual views contrastive constraints are designed to enforce the robustness of transferable structural knowledge. To the best of our knowledge, our work makes one of the first attempts to generalize the learning of hyper-relation structures, offering high flexibility and ease of use without reliance on any external resources. HyRel demonstrates SOTA performance compared to existing methods under extensive inductive settings, particularly on fully unseen graphs, and validates the efficacy of learning hyper-relation structures for improving generalization. The code is available online at https://github.com/hncps6/HyRel.

Primary Subject Area: [Content] Media Interpretation

Relevance To Conference:

This paper contributes to the advancement of multimedia/multimodal processing by proposing a novel approach to enhance its generalization capability.
The various elements within multimedia data or social networks, such as images, videos, texts, users, etc., and their interrelationships naturally form complex graphs. This paper proposes an inductive link prediction method for inferring missing link in unseen graphs, such as predicting absent links in social networks. This paper offers the method to accurately reason on new, unseen multimedia data, demonstrating strong adaptability across various multimedia scenarios, thereby providing robust support for multimedia interpretation.
Such an approach enables more accurate and comprehensive understanding and processing of multimedia data, thereby providing powerful tools and methods for interpreting, analyzing, and generating multimedia content. Consequently, this work offers new insights and techniques for the multimedia/multimodal processing field, driving its progress and development.

Submission Number: 5724