Inductive Reasoning for Temporal Knowledge Graphs with Emerging Entities
Summary: arXiv:2604.10164v1 Announce Type: new
Abstract: Reasoning on Temporal Knowledge Graphs (TKGs) is essential for predicting future events and time-aware facts. While existing methods are effective at capturing relational dynamics, their performance is limited by a closed-world assumption, which fails to account for emerging entities not present in the training. Notably, these entities continuously join the network without historical interactions. Empirical study reveals that emerging entities are widespread in TKGs, comprising roughly 25% of all entities. The absence of historical interactions of these entities leads to significant performance degradation in reasoning tasks.
Whereas, we observe that entities with semantic similarities often exhibit comparable interaction histories, suggesting the presence of transferable temporal patterns. Inspired by this insight, we propose TransFIR (Transferable Inductive Reasoning), a novel framework that leverages historical interaction sequences from semantically similar known entities to support inductive reasoning. Specifically, we propose a codebook-based classifier that categorizes emerging entities into latent semantic clusters, allowing them to adopt reasoning patterns from similar entities.
Key Insights and Methodology
The research highlights several key insights and methodologies that contribute to the development of TransFIR:
- Closed-World Assumption Limitation: Traditional methods often assume a closed-world scenario, neglecting the impact of emerging entities.
- Empirical Findings: Emerging entities constitute approximately 25% of all entities within Temporal Knowledge Graphs, significantly affecting reasoning performance.
- Transferable Temporal Patterns: Entities that share semantic similarities display analogous interaction histories, providing a basis for knowledge transfer.
- Codebook-Based Classifier: The proposed framework uses a classifier to categorize emerging entities into meaningful semantic clusters, enhancing inductive reasoning capabilities.
Experimental Results
The effectiveness of TransFIR was rigorously tested against existing baseline methods. The results indicate a remarkable improvement in reasoning tasks involving emerging entities. Key findings include:
- TransFIR achieved an average improvement of 28.6% in Mean Reciprocal Rank (MRR) across multiple datasets.
- The incorporation of historical interaction sequences from similar entities significantly enhanced the performance of reasoning tasks.
- TransFIR outperformed all baseline models, demonstrating its superior capability in handling emerging entities.
Conclusion and Future Work
The introduction of TransFIR marks a significant advancement in the field of Temporal Knowledge Graphs, addressing critical limitations of existing approaches through innovative reasoning techniques. The research opens avenues for future studies aimed at improving reasoning accuracy and efficiency in dynamic networks.
The full implementation of TransFIR is available for public access at GitHub Repository, encouraging further exploration and development in this promising area of study.