Deep embeddings and Graph Neural Networks: using context to improve domain-independent predictions
Graph neural networks (GNNs) are deep learning architectures that apply graph convolutions through message-passing processes between nodes, represented as embeddings. GNNs have recently become popular because of their ability to obtain a contextual representation of each node taking into account inf...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión publicada |
| Fecha de publicación: | 2023 |
| País: | España |
| Institución: | Universidad de Sevilla (US) |
| Repositorio: | idUS. Depósito de Investigación de la Universidad de Sevilla |
| OAI Identifier: | oai:idus.us.es:11441/147545 |
| Acceso en línea: | https://hdl.handle.net/11441/147545 https://doi.org/10.1007/s10489-023-04685-3 |
| Access Level: | acceso abierto |
| Palabra clave: | Knowledge graphs Graph neural networks Attributive embeddings Deep graph embeddings Machine learning |
| Sumario: | Graph neural networks (GNNs) are deep learning architectures that apply graph convolutions through message-passing processes between nodes, represented as embeddings. GNNs have recently become popular because of their ability to obtain a contextual representation of each node taking into account information from its surroundings. However, existing work has focused on the development of GNN architectures, using basic domain-specific information about the nodes to compute embeddings. Meanwhile, in the closely-related area of knowledge graphs, much effort has been put towards developing deep learning techniques to obtain node embeddings that preserve information about relationships and structure without relying on domain-specific data. The potential application of deep embeddings of knowledge graphs in GNNs remains largely unexplored. In this paper, we carry out a number of experiments to answer open research questions about the impact on GNNs performance when combined with deep embeddings. We test 7 different deep embeddings across several attribute prediction tasks in two state-of-art attribute-rich datasets. We conclude that, while there is a significant performance improvement, its magnitude varies heavily depending on the specific task and deep embedding technique considered. |
|---|