Structural Alignment in Link Prediction
Jeffrey Seathrún Sardina
TL;DR
The work questions the dominance of embedding-based KGEMs for link prediction and proposes a structure-first alternative grounded in triple-level graph structure. It formalizes the Structural Alignment Hypothesis and develops TWIG to simulate KGEM outputs from KG structure, while TWIG-I directly performs structure-based LP without embeddings. Empirical results show that structure can both predict KGEM hyperparameter preferences and enable competitive or superior LP performance, with demonstrated cross-KG and cross-domain transfer capabilities. The findings suggest structure as a viable, scalable basis for KG learning, with implications for KG construction, transfer learning, and future graph-foundation models. Overall, the thesis provides a cohesive framework and open-source tools for structure-driven KG learning and cross-graph LP, offering a principled alternative to embedding-centric approaches.
Abstract
While Knowledge Graphs (KGs) have become increasingly popular across various scientific disciplines for their ability to model and interlink huge quantities of data, essentially all real-world KGs are known to be incomplete. As such, with the growth of KG use has been a concurrent development of machine learning tools designed to predict missing information in KGs, which is referred to as the Link Prediction Task. The majority of state-of-the-art link predictors to date have followed an embedding-based paradigm. In this paradigm, it is assumed that the information content of a KG is best represented by the (individual) vector representations of its nodes and edges, and that therefore node and edge embeddings are particularly well-suited to performing link prediction. This thesis proposes an alternative perspective on the field's approach to link prediction and KG data modelling. Specifically, this work re-analyses KGs and state-of-the-art link predictors from a graph-structure-first perspective that models the information content of a KG in terms of whole triples, rather than individual nodes and edges. Following a literature review and two core sets of experiments, this thesis concludes that a structure-first perspective on KGs and link prediction is both viable and useful for understanding KG learning and for enabling cross-KG transfer learning for the link prediction task. This observation is used to create and propose the Structural Alignment Hypothesis, which postulates that link prediction can be understood and modelled as a structural task. All code and data used for this thesis are open-sourced. This thesis was written bilingually, with the main document in English and an informal extended summary in Irish. An Irish-language translation dictionary of machine learning terms (the Foclóir Tráchtais) created for this work is open-sourced as well.