AdaTKG: Adaptive Memory for Temporal Knowledge Graph Reasoning

2026-05-07

AI Summary

Existing approaches generate static entity representations that struggle to capture the dynamic evolution of entities in temporal knowledge graphs. To address this limitation, this work proposes AdaTKG, a novel framework that employs an adaptive memory module to update entity representations online, effectively modeling their semantic evolution over time. AdaTKG introduces a lightweight, shared, learnable exponential moving average mechanism that avoids maintaining separate parameters for each entity, thereby enabling generalization to entities unseen during training. Experimental results demonstrate that AdaTKG significantly outperforms state-of-the-art methods across multiple benchmark datasets, confirming the effectiveness and superiority of the proposed adaptive memory mechanism for temporal reasoning.

Abstract

Temporal knowledge graphs (TKGs) represent time-stamped relational facts and support a wide range of reasoning tasks over evolving events. However, existing methods produce entity representations that are static at the entity level, in that each representation is a function of learned parameters only and retains no trace of the interactions in which the entity has participated. In this paper, we depart from this static view and propose that each entity be modeled as an adaptive process whose representation is refined every time the entity participates in a fact. To this end, we propose AdaTKG, which maintains a per-entity memory that is updated with every observed interaction, with the memory accumulating online and predictions improving as more interactions arrive. Specifically, we instantiate the memory update as a learnable exponential moving average governed by a single shared scalar instead of using learnable parameters for each entity, enabling AdaTKG to handle entities unseen during training. Extensive experiments confirm consistent gains over TKG baselines, demonstrating the effectiveness of adaptive memory. Code is publicly available at: https://github.com/seunghan96/AdaTKG.

Problem

• Temporal Knowledge Graph
• Entity Representation
• Static Representation
• Dynamic Modeling
• Reasoning

Innovation

• adaptive memory
• temporal knowledge graph
• entity representation
• exponential moving average
• inductive reasoning