Abstract:In practical applications, the temporal completeness of knowledge graphs is of great importance. However, previous studies have mostly focused on static knowledge graphs, generally neglecting the dynamic evolutionary properties of facts. Moreover, the unpredictable and limited availability of temporal knowledge graphs, together with the complex temporal dependency patterns, make current models inadequate for effectively describing facts that experience temporal transitions. To better represent the evolution of things over time, we provide a learning technique that uses quaternion rotation to describe temporal knowledge graphs. This technique describes the evolution of entities as a temporal rotation change in quaternion space. Compared to the Ermitian inner product in complex number space, the Hamiltonian product in quaternion space is better at showing how things might be connected. This leads to a learning process that is both more effective and more articulate. Experimental results demonstrate that our learning method significantly outperforms existing methods in capturing the dynamic evolution of temporal knowledge graphs, with improved accuracy and robustness across a range of benchmark datasets.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is that most of the existing knowledge graph research focuses on static knowledge graphs and ignores the dynamic evolution characteristics of facts. Specifically, the article points out: 1. **Limitations of Static Knowledge Graphs**: Most existing research mainly focuses on static knowledge graphs and ignores time attributes and the changes of facts over time. This leads to poor performance of existing models when dealing with knowledge graphs containing time information. 2. **Complexity of Time - Dependent Patterns**: The time - dependent patterns in time knowledge graphs are complex and unpredictable, and existing models are difficult to effectively describe facts undergoing time transitions. 3. **Data Sparsity and Incompleteness**: The data of time knowledge graphs often has the challenges of sparsity and incompleteness, making accurate modeling more difficult. To solve these problems, the author proposes a learning technique based on quaternion rotation (TKGR) to more effectively represent and capture the time evolution of entities in time knowledge graphs. This method describes the time changes of entities through rotation transformations in the quaternion space, and uses the Hamiltonian product to better show the potential connections between things than the Ermitian inner product in the complex space. ### Main Contributions - **Introduction of Quaternion Rotation**: Use quaternion rotation to describe time knowledge graphs and improve the effectiveness and expressiveness of the learning process. - **Application of Hamiltonian Product**: Use the Hamiltonian product of quaternions to accurately describe time and relationship changes in the knowledge network. - **Experimental Verification**: Through comparative experiments with existing methods, it proves the superior performance of the proposed method on multiple benchmark data sets and shows its advantages in capturing the dynamic evolution of time knowledge graphs. ### Key Issues in the Abstract - **Importance of Time Attributes**: In practical applications, the time integrity of time knowledge graphs is crucial. - **Deficiencies of Existing Methods**: Existing methods cannot effectively describe facts undergoing time transitions, especially when dealing with complex time - dependent patterns. - **Solution**: Propose a learning technique based on quaternion rotation to better represent the evolution of things over time. Through these improvements, this research aims to enhance the representation and reasoning abilities of time knowledge graphs, thereby providing more accurate and robust results in various application scenarios.

Revolutionary Strategy for Depicting Knowledge Graphs with Temporal Attributes

RotateQVS: Representing Temporal Information as Rotations in Quaternion Vector Space for Temporal Knowledge Graph Completion

EHPR: Learning Evolutionary Hierarchy Perception Representation based on Quaternion for Temporal Knowledge Graph Completion

Temporal Knowledge Graph Completion with Time-sensitive Relations in Hypercomplex Space

Know-Evolve: Deep Temporal Reasoning for Dynamic Knowledge Graphs

TeRo: A Time-aware Knowledge Graph Embedding via Temporal Rotation

ChronoR: Rotation Based Temporal Knowledge Graph Embedding

Learning Dynamic Embeddings for Temporal Knowledge Graphs

Dynamic dual quaternion knowledge graph embedding

Householder Transformation-Based Temporal Knowledge Graph Reasoning

Distance-Adaptive Quaternion Knowledge Graph Embedding with Bidirectional Rotation

Temporal Knowledge Graph Reasoning Based on Evolutional Representation and Contrastive Learning

Temporal knowledge graph representation learning with local and global evolutions

Learning Hierarchy-Aware Quaternion Knowledge Graph Embeddings with Representing Relations as 3D Rotations.

Bending the Future: Autoregressive Modeling of Temporal Knowledge Graphs in Curvature-Variable Hyperbolic Spaces

An embedding model for temporal knowledge graphs with long and irregular intervals

Temporal Knowledge Graph Reasoning Based on Dynamic Fusion Representation Learning

A Survey on Temporal Knowledge Graph: Representation Learning and Applications

Transformer-based Reasoning for Learning Evolutionary Chain of Events on Temporal Knowledge Graph

A knowledge representation model based on the geographic spatiotemporal process