Knowledge Graph Completion with Relation-Aware Anchor Enhancement

TL;DR

The paper addresses text-based knowledge graph completion by leveraging relation-aware anchors to supply tail-context during link prediction. It introduces the RAA-KGC framework, which generates relation-aware anchors from a head's neighborhood, constructs anchor-enhanced queries, and trains with a contrastive objective using two encoders. Empirically, RAA-KGC yields significant gains over strong baselines on WN18RR, Wikidata5M-Trans, and in inductive settings, while remaining compatible with existing KGC models. The work demonstrates that relation-aware anchor information can markedly improve discriminability with minimal architectural changes, offering practical improvements for scalable KGC.

Abstract

Text-based knowledge graph completion methods take advantage of pre-trained language models (PLM) to enhance intrinsic semantic connections of raw triplets with detailed text descriptions. Typical methods in this branch map an input query (textual descriptions associated with an entity and a relation) and its candidate entities into feature vectors, respectively, and then maximize the probability of valid triples. These methods are gaining promising performance and increasing attention for the rapid development of large language models. According to the property of the language models, the more related and specific context information the input query provides, the more discriminative the resultant embedding will be. In this paper, through observation and validation, we find a neglected fact that the relation-aware neighbors of the head entities in queries could act as effective contexts for more precise link prediction. Driven by this finding, we propose a relation-aware anchor enhanced knowledge graph completion method (RAA-KGC). Specifically, in our method, to provide a reference of what might the target entity be like, we first generate anchor entities within the relation-aware neighborhood of the head entity. Then, by pulling the query embedding towards the neighborhoods of the anchors, it is tuned to be more discriminative for target entity matching. The results of our extensive experiments not only validate the efficacy of RAA-KGC but also reveal that by integrating our relation-aware anchor enhancement strategy, the performance of current leading methods can be notably enhanced without substantial modifications.

Figures (4)

• Figure 1: In this figure, the yellow bricks denote relation-aware entities which share the same head entity and relation in the testing triple. These entities share important concept information with the target entity (the dashed blue brick), thus can be used to guide for better link prediction.
• Figure 2: Framework illustration of the proposed RAA-KGC. In this figure, the yellow bricks denote relation-aware entities that share the same head entity and relation. The dashed blue brick denotes the target entity. The core idea of RAA-KGC is to generate a general example of what might the tail (or head) entity be like by using relation-aware entities, and then pull its embedding towards the neighborhood of the anchor.
• Figure 3: Visualization of entity embeddings with the different head entities and relations using t-SNE in NT and IRT scenarios. Points in same color indicate that their head entities and relations are the same. $h_1$, $h_2$, $h_3$ represent different query head entities, $r_1$ and $r_2$ indicate knowledge graph relations. The NMI of (a) and (b) are 0.75 and 0.62, respectively.
• Figure 4: The difference value of Hit@1 between RAA-KGC and SimKGC on different relations in the WN18RR dataset.