CensNet: Convolution with Edge-Node Switching in Graph Neural Networks

Jiang, Xiaoling; Ji, Pengsheng; Li, Sheng

doi:10.24963/ijcai.2019/369

Cited by 65 publications

(31 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…connected or not) of relation, and only focus on entity embedding learning. Our work is related to graph neural networks, such as the graph convolutional networks (GCN) (Kipf and Welling, 2017) and its variants (Wu et al, 2020;Jiang et al, 2019Jiang et al, , 2020, which learn node embeddings by feature propagation. In the following, we mainly review the most relevant works in two aspects, i.e., graph embedding learning with external text and knowledge graph construction.…”

Section: Related Workmentioning

confidence: 99%

Edge: Enriching Knowledge Graph Embeddings with External Text

Rezayi¹,

Zhao²,

Kim³

et al. 2021

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua

Self Cite

View full text Add to dashboard Cite

Knowledge graphs suffer from sparsity which degrades the quality of representations generated by various methods. While there is an abundance of textual information throughout the web and many existing knowledge bases, aligning information across these diverse data sources remains a challenge in the literature. Previous work has partially addressed this issue by enriching knowledge graph entities based on "hard" co-occurrence of words present in the entities of the knowledge graphs and external text, while we achieve "soft" augmentation by proposing a knowledge graph enrichment and embedding framework named EDGE. Given an original knowledge graph, we first generate a rich but noisy augmented graph using external texts in semantic and structural level. To distill the relevant knowledge and suppress the introduced noise, we design a graph alignment term in a shared embedding space between the original and augmented graph. To enhance the embedding learning on the augmented graph, we further regularize the locality relationship of target entity based on negative sampling. Experimental results on four benchmark datasets demonstrate the robustness and effectiveness of EDGE in link prediction and node classification.

show abstract

Section: Related Workmentioning

confidence: 99%

Edge: Enriching Knowledge Graph Embeddings with External Text

Rezayi¹,

Zhao²,

Kim³

et al. 2021

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langua

Self Cite

View full text Add to dashboard Cite

show abstract

“…generate the path representation between target nodes. CensNet (Jiang, Ji, and Li 2019) uses graph convolution network with edge-node switching without the interaction of edges, which is hard to capture topological structure information through paths. GAS…”

Section: Graph Neural Networkmentioning

confidence: 99%

Attributed Graph Neural Networks for Recommendation Systems on Large-Scale and Sparse Graph

Zhang¹,

Ou²,

Zhang³

et al. 2021

Preprint

View full text Add to dashboard Cite

Link prediction in structured-data is an important problem for many applications, especially for recommendation systems. Existing methods focus on how to learn the node representation based on graph-based structure. High-dimensional sparse edge features are not fully exploited. Because balancing precision and computation efficiency is significant for recommendation systems in real world, multiple-level feature representation in large-scale sparse graph still lacks effective and efficient solution. In this paper, we propose a practical solution about graph neural networks called Attributed Graph Convolutional Networks(AGCN) to incorporate edge attributes when apply graph neural networks in largescale sparse networks. We formulate the link prediction problem as a subgraph classification problem. We firstly propose an efficient two-level projection to decompose topological structures to node-edge pairs and project them into the same interaction feature space. Then we apply multi-layer GCN to combine the projected nodeedge pairs to capture the topological structures. Finally, the pooling representation of two units is treated as the input of classifier to predict the probability. We conduct offline experiments on two industrial datasets and one public dataset and demonstrate that AGCN outperforms other excellent baselines. Moreover, we also deploy AGCN method to important scenarios on Xianyu and AliExpress. In online systems, AGCN achieves over 5% improvement on online metrics.

show abstract

“…Recently, increasing manually-designed GNNs [10,19,25,46] try to incorporate edge features into models for better exploiting relation information. Specifically, they stack the building layers in a sequential architecture and iteratively learn features of entity and edge.…”

Section: Edge-updating Graphmentioning

confidence: 99%

“…Exploiting Edge Features. Recently, researchers [10,19,25,38,46] have tried to intergrate edge features into GNN architecture. Schlichtkrull et al [38] propose R-GCNs to model relational data by grouping edges on graph, which indicates the edges cannot include continuous attributes.…”

Section: Related Workmentioning

confidence: 99%

“…Schlichtkrull et al [38] propose R-GCNs to model relational data by grouping edges on graph, which indicates the edges cannot include continuous attributes. Works [10,19,46] assign high-dimensional feature for edges and iteratively update edge features using the same way as entity updating. These methods take no account of the dependencies between multi-level relation features and lack of diversified update functions to compute the edge features.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Edge-featured Graph Neural Architecture Search

Cai¹,

Zhang²,

Han³

et al. 2021

Preprint

View full text Add to dashboard Cite

Graph neural networks (GNNs) have been successfully applied to learning representation on graphs in many relational tasks. Recently, researchers study neural architecture search (NAS) to reduce the dependence of human expertise and explore better GNN architectures, but they over-emphasize entity features and ignore latent relation information concealed in the edges. To solve this problem, we incorporate edge features into graph search space and propose Edge-featured Graph Neural Architecture Search (EGNAS) to find the optimal GNN architecture. Specifically, we design rich entity and edge updating operations to learn high-order representations, which convey more generic message passing mechanisms. Moreover, the architecture topology in our search space allows to explore complex feature dependence of both entities and edges, which can be efficiently optimized by differentiable search strategy. Experiments at three graph tasks on six datasets show EGNAS can search better GNNs with higher performance than current state-of-the-art human-designed and searched-based GNNs. 2 * Corresponding author. 2 Codes have been provided in supplementary materials, and will be released via GitHub.Preprint. Under review.

show abstract

CensNet: Convolution with Edge-Node Switching in Graph Neural Networks

Cited by 65 publications

References 8 publications

Edge: Enriching Knowledge Graph Embeddings with External Text

Edge: Enriching Knowledge Graph Embeddings with External Text

Attributed Graph Neural Networks for Recommendation Systems on Large-Scale and Sparse Graph

Edge-featured Graph Neural Architecture Search

Contact Info

Product

Resources

About