Chenxi Wan, Xunkai Li, Yilong Zuo, Haokun Deng, Sihan Li, Bowen Fan, Hongchao Qin, Ronghua Li, Guoren Wang
OpenMAG is a comprehensive benchmark designed to evaluate Multimodal-Attributed Graph (MAG) models across various domains, encoders, and tasks, addressing existing limitations in fair evaluation standards.
OpenMAG is a new tool to help researchers evaluate complex models that combine graph structures with multiple types of data. These models are important for understanding real-world systems, like social networks or biological systems, where different types of information need to be integrated. OpenMAG includes a wide range of datasets and models to provide a fair and thorough testing ground. This helps ensure that new models are evaluated consistently and effectively, leading to better advancements in the field.
