TRIX: A More Expressive Model for Zero-shot Domain Transfer in Knowledge Graphs

2025-02-26

AI Summary

Zero-shot cross-domain knowledge graph completion (KGC) suffers from weak expressiveness and difficulty in jointly predicting unseen entities and relations. Method: We propose the first fully inductive KGC framework, featuring a unified triplet embedding architecture with theoretically superior expressiveness over existing models, and an inductive graph neural network based on high-order tensor interaction and learnable structural encoding—requiring no target-domain prior knowledge or fine-tuning. Contribution/Results: This work achieves, for the first time, end-to-end joint modeling of entities and relations under fully inductive settings. It significantly improves zero-shot generalization, outperforming state-of-the-art fully inductive models across multiple benchmarks, and surpasses large-context language models on cross-domain prediction tasks. The code is publicly available.

Technology Category

• Data Mining & Knowledge Management: Linked Open Data, Knowledge Graphs & KB Completion
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Machine Learning: Graph-based Machine Learning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Fully inductive knowledge graph models can be trained on multiple domains and subsequently perform zero-shot knowledge graph completion (KGC) in new unseen domains. This is an important capability towards the goal of having foundation models for knowledge graphs. In this work, we introduce a more expressive and capable fully inductive model, dubbed TRIX, which not only yields strictly more expressive triplet embeddings (head entity, relation, tail entity) compared to state-of-the-art methods, but also introduces a new capability: directly handling both entity and relation prediction tasks in inductive settings. Empirically, we show that TRIX outperforms the state-of-the-art fully inductive models in zero-shot entity and relation predictions in new domains, and outperforms large-context LLMs in out-of-domain predictions. The source code is available at https://github.com/yuchengz99/TRIX.

Problem

• Zero-shot domain transfer in knowledge graphs
• Enhanced triplet embeddings for KGC
• Entity and relation prediction in new domains

Innovation

• Expressive triplet embeddings generation
• Handles entity and relation prediction
• Outperforms in zero-shot domain transfer