翻译：本文提出了一种基于多模态特征加强的实体对齐模型其在传统模型仅使用各模态编码器对各模态输入进行编码以初步获取实体表示的基础上通过多模态预训练模型、OCR模型、GATv2网络对图像与结构特征进行增强并额外利用实体模态信息分布提高模型对实体各属性的理解与联合建模能力其整体架构图如图所示。

This article proposes a multi-modal feature-enhanced entity alignment model, which enhances image and structural features with multi-modal pre-training models, OCR models, and GATv2 networks on the basis of traditional models that only use each modal encoder to encode each modal input to initially obtain entity representations. Additionally, the model utilizes the distribution of entity modal information to improve the understanding and joint modeling ability of the model for various entity attributes. The overall architecture of the model is shown in the figure.