MarKG

该数据集名为MarKG，是一个专为支持多模态类比推理任务而设计的多模态知识图谱。它提供了关键的结构知识，这些知识能够增强实体和关系的表示学习，这对于类比推理任务至关重要。该数据集旨在研究和探索在多模态类比推理中的泛化与迁移问题。

This dataset, named MarKG, is a multimodal knowledge graph specifically designed to support multimodal analogical reasoning tasks. It provides critical structural knowledge that enhances representation learning of entities and relations, which is indispensable for analogical reasoning tasks. This dataset is intended to investigate and explore the generalization and transfer issues in multimodal analogical reasoning.

搜集汇总

背景与挑战

背景概述

MKG_Analogy是一个用于多模态类比推理任务的数据集，包含知识图谱数据MarKG和类比推理数据MARS，支持多种基准方法的评估和训练。