![[2510.01510] Flock: A Knowledge Graph Foundation Model via Learning on Random Walks](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2510.01510] Flock: A Knowledge Graph Foundation Model via Learning on Random Walks
Summary
The paper presents Flock, a knowledge graph foundation model that enhances zero-shot link prediction by employing probabilistic node-relation equivariance, improving performance on diverse knowledge graphs.
Why It Matters
Flock addresses significant limitations in existing knowledge graph foundation models by introducing a probabilistic approach to equivariance. This innovation allows for better generalization to novel entities and relations, which is crucial for advancing applications in AI that rely on knowledge graphs, such as recommendation systems and semantic search.
Key Takeaways
- Flock utilizes probabilistic node-relation equivariance to enhance model expressiveness.
- The model achieves state-of-the-art performance on entity and relation prediction tasks.
- Flock is a universal approximator for isomorphism-invariant link-level functions.
- The new diagnostic dataset Petals reveals limitations of current models, which Flock overcomes.
- Code for Flock is publicly available, promoting further research and application.
Computer Science > Machine Learning arXiv:2510.01510 (cs) [Submitted on 1 Oct 2025 (v1), last revised 17 Feb 2026 (this version, v2)] Title:Flock: A Knowledge Graph Foundation Model via Learning on Random Walks Authors:Jinwoo Kim, Xingyue Huang, Krzysztof Olejniczak, Kyungbin Min, Michael Bronstein, Seunghoon Hong, İsmail İlkan Ceylan View a PDF of the paper titled Flock: A Knowledge Graph Foundation Model via Learning on Random Walks, by Jinwoo Kim and 6 other authors View PDF HTML (experimental) Abstract:We study the problem of zero-shot link prediction on knowledge graphs (KGs), which requires models to generalize to novel entities and novel relations. Knowledge graph foundation models (KGFMs) address this task by enforcing equivariance over both nodes and relations, which enables them to learn structural properties of nodes and relations that transfer to novel KGs with similar structure. However, the conventional notion of deterministic equivariance inherently limits the expressive power of KGFMs, as it prevents them from distinguishing relations that are structurally similar but semantically distinct. To overcome this limitation, we propose to leverage probabilistic node-relation equivariance, which preserves equivariance in distribution while using structured randomness to break symmetries at inference time. Building on this principle, we present Flock, a KGFM that iteratively samples random walks, encodes them into sequences, embeds them with a sequence model, and a...