analiza aiqbench
This research demonstrates that standard Graph Neural Networks can achieve zero-shot transfer across 40 unseen benchmarks by reifying graph structures into a fixed vocabulary, effectively matching the performance of specialized foundation models like ULTRA without architectural hard-coding. By treating relation types as shared nodes rather than learnable parameters, the method simplifies knowledge graph reasoning and extends applicability to relational databases using only schema information. This approach offers a significant efficiency gain for inductive link prediction tasks by leveraging off-the-shelf GNNs trained on minimal data.
Reification as a Transferable Vocabulary: Zero-Shot Link Prediction with Vanilla GNNs
arXiv:2609.11347v1 Announce Type: cross Abstract: Knowledge graph foundation models such as ULTRA achieve zero-shot link prediction on unseen graphs through dedicated architectures that hard-code a transfer mechanism. In this work we move that mechanism out of the architecture and into the representation, by \emph{reifying} the input graph: every fact becomes a node, connected to its subject, object, and relation type through a fixed vocabulary of six meta-relations, with relation types as anonymous shared nodes rather than model parameters. On this representation, five textbook GNNs (GAT, GINE with sum and with mean+max aggregation, GraphSAGE, R-GCN), each trained on a single knowledge graph of 4,245 triples for 30 minutes on one NVIDIA A100, transfer zero-shot to 40 inductive link-prediction benchmarks. The best of them, an off-the-shelf GAT, matches ULTRA, a dedicated foundation model pretrained on three graphs, across ULTRA's own evaluation suite. The same fixed vocabulary extends to relational databases, a row becoming an entity and a foreign-key column a relation type; a preliminary probe on two unseen databases, with no cell values, schema text or in-context labels, shows a model of this family pretrained on three knowledge graphs ranking foreign-key targets far above random-initialization and degree controls. We release the code, the checkpoints, and the evaluation pipeline for all 40 benchmarks.
Abstract:Knowledge graph foundation models such as ULTRA achieve zero-shot link prediction on unseen graphs through dedicated architectures that hard-code a transfer mechanism. In this work we move that mechanism out of the architecture and into the representation, by \emph{reifying} the input graph: every fact becomes a node, connected to its subject, object, and relation type through a fixed vocabulary of six meta-relations, with relation types as anonymous shared nodes rather than model parameters. On this representation, five textbook GNNs (GAT, GINE with sum and with mean+max aggregation, GraphSAGE, R-GCN), each trained on a single knowledge graph of 4,245 triples for 30 minutes on one NVIDIA A100, transfer zero-shot to 40 inductive link-prediction benchmarks. The best of them, an off-the-shelf GAT, matches ULTRA, a dedicated foundation model pretrained on three graphs, across ULTRA's own evaluation suite. The same fixed vocabulary extends to relational databases, a row becoming an entity and a foreign-key column a relation type; a preliminary probe on two unseen databases, with no cell values, schema text or in-context labels, shows a model of this family pretrained on three knowledge graphs ranking foreign-key targets far above random-initialization and degree controls. We release the code, the checkpoints, and the evaluation pipeline for all 40 benchmarks.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2609.11347 [cs.LG]
(or arXiv:2609.11347v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.11347 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 302 cuvinte
Subiecte conexe
The Universal Classifier for Graph Learning
arXiv · 30 sept. 2026
[AINews] not much happened today
Latent Space · 3 oct. 2026
Adaptation Interfaces for In-Context Tabular Foundation Models in Time-to-Event Prediction
arXiv · 7 sept. 2026
The Entangled Dimensions of Decision Theory
LessWrong · 30 iul. 2026