翻译：大多早期实体对齐模型使用图卷积网络或者效果更佳的图注意力网络对实体的关系三元组进行结构建模但应用于实体对齐任务时传统静态图注意力网络的局限性主要体现在共享权重矩阵与静态性两方面。

Most early entity alignment models use graph convolutional networks or more effective graph attention networks to structurally model the relationship triples of entities. However, when applied to entity alignment tasks, the limitations of traditional static graph attention networks mainly lie in the aspects of weight matrix sharing and static nature.