图表示学习侧重于图结构的处理。知识图谱嵌入模型和规则学习等更加侧重于语义和逻辑结构特征的学习,更好的知识图谱表示学习方法需要综合利用好语义,逻辑结构的特征学习和图结构的特征学习等多种方法
- RGCN: Modeling Relational Data with Graph Convolutional Networks. (ESWC 2018).
- COMPGCN: Composition-based Multi-Relational Graph Convolutional Networks. (ICLR 2020).
Knowledge transfer for out-of-knowledge-base entities: a graph neural network approach. (IJCAI 2017). - Long-tail Relation Extraction via Knowledge Graph Embeddings and Graph Convolution Networks. (NAACL 2019).
- RDGCN: Relation-Aware Entity Alignment for Heterogeneous Knowledge Graphs. (IJCAI 2019).
- Commonsense Knowledge Aware Conversation Generation with Graph Attention. (IJCAI 2018).
- End-to-end Structure-Aware Convolutional Networks for Knowledge Base Completion. (AAAI 2019).
GCN 和 ConvE 的结合,整体结构是encoder-decoder。带权重的GCN学习entity 表示,TransE学习relation 表示。将entity 表示和relation表示作为encoder的中间结果,作为decoder 的输入。