Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNs

Knowledge graph embedding has been identified as an effective method for node-level classification tasks in directed graphs, the objective of which is to ensure that nodes of different categories are embedded as far apart as possible in the feature space. The directed graph is a general representati...

Full description

Saved in:
Bibliographic Details
Main Authors: Guoqiang Hou, Qiwen Yu, Fan Chen, Guang Chen
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Mathematics
Subjects:
Online Access:https://www.mdpi.com/2227-7390/12/23/3689
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850107053209550848
author Guoqiang Hou
Qiwen Yu
Fan Chen
Guang Chen
author_facet Guoqiang Hou
Qiwen Yu
Fan Chen
Guang Chen
author_sort Guoqiang Hou
collection DOAJ
description Knowledge graph embedding has been identified as an effective method for node-level classification tasks in directed graphs, the objective of which is to ensure that nodes of different categories are embedded as far apart as possible in the feature space. The directed graph is a general representation of unstructured knowledge graphs. However, existing methods lack the ability to simultaneously approximate high-order filters and globally pay attention to the task-related connectivity between distant nodes for directed graphs. To address this limitation, a directed spectral graph transformer (DSGT), a hybrid architecture model, is constructed by integrating the graph transformer and directed spectral graph convolution networks. The graph transformer leverages multi-head attention mechanisms to capture the global connectivity of the feature graph from different perspectives in the spatial domain, which bridges the gap between frequency responses and, further, naturally couples the graph transformer and directed graph convolutional neural networks (GCNs). In addition to the inherent hard inductive bias of DSGT, we introduce directed node positional and structure-aware edge embedding to provide topological prior knowledge. Extensive experiments demonstrate that the DSGT exhibits state-of-the-art (SOTA) or competitive node-level representation capabilities across datasets of varying attributes and scales. Furthermore, the experimental results indicate that the homophily and degree of correlation of the nodes significantly influence the classification performance of the model. This finding opens significant avenues for future research.
format Article
id doaj-art-98a01b4c5fa54caeb1f9c95f54adb71b
institution OA Journals
issn 2227-7390
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Mathematics
spelling doaj-art-98a01b4c5fa54caeb1f9c95f54adb71b2025-08-20T02:38:41ZengMDPI AGMathematics2227-73902024-11-011223368910.3390/math12233689Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNsGuoqiang Hou0Qiwen Yu1Fan Chen2Guang Chen3College of Intelligent Systems Science and Engineering, Harbin Engineering University, Harbin 150001, ChinaCollege of Intelligent Systems Science and Engineering, Harbin Engineering University, Harbin 150001, ChinaCollege of Intelligent Manufacturing, Chongqing Vocational and Technical College of Industry and Trade, Chongqing 401120, ChinaCollege of Intelligent Systems Science and Engineering, Harbin Engineering University, Harbin 150001, ChinaKnowledge graph embedding has been identified as an effective method for node-level classification tasks in directed graphs, the objective of which is to ensure that nodes of different categories are embedded as far apart as possible in the feature space. The directed graph is a general representation of unstructured knowledge graphs. However, existing methods lack the ability to simultaneously approximate high-order filters and globally pay attention to the task-related connectivity between distant nodes for directed graphs. To address this limitation, a directed spectral graph transformer (DSGT), a hybrid architecture model, is constructed by integrating the graph transformer and directed spectral graph convolution networks. The graph transformer leverages multi-head attention mechanisms to capture the global connectivity of the feature graph from different perspectives in the spatial domain, which bridges the gap between frequency responses and, further, naturally couples the graph transformer and directed graph convolutional neural networks (GCNs). In addition to the inherent hard inductive bias of DSGT, we introduce directed node positional and structure-aware edge embedding to provide topological prior knowledge. Extensive experiments demonstrate that the DSGT exhibits state-of-the-art (SOTA) or competitive node-level representation capabilities across datasets of varying attributes and scales. Furthermore, the experimental results indicate that the homophily and degree of correlation of the nodes significantly influence the classification performance of the model. This finding opens significant avenues for future research.https://www.mdpi.com/2227-7390/12/23/3689knowledge graph embeddingshybrid architecturegraph transformerdirected spectral graph convolution networksnode-level representation learning
spellingShingle Guoqiang Hou
Qiwen Yu
Fan Chen
Guang Chen
Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNs
Mathematics
knowledge graph embeddings
hybrid architecture
graph transformer
directed spectral graph convolution networks
node-level representation learning
title Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNs
title_full Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNs
title_fullStr Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNs
title_full_unstemmed Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNs
title_short Directed Knowledge Graph Embedding Using a Hybrid Architecture of Spatial and Spectral GNNs
title_sort directed knowledge graph embedding using a hybrid architecture of spatial and spectral gnns
topic knowledge graph embeddings
hybrid architecture
graph transformer
directed spectral graph convolution networks
node-level representation learning
url https://www.mdpi.com/2227-7390/12/23/3689
work_keys_str_mv AT guoqianghou directedknowledgegraphembeddingusingahybridarchitectureofspatialandspectralgnns
AT qiwenyu directedknowledgegraphembeddingusingahybridarchitectureofspatialandspectralgnns
AT fanchen directedknowledgegraphembeddingusingahybridarchitectureofspatialandspectralgnns
AT guangchen directedknowledgegraphembeddingusingahybridarchitectureofspatialandspectralgnns