MAPLE

The MAPLE benchmark constructed by us contains 20 datasets across 19 fields for scientific literature tagging. It also has a graph format, which can be used for graph mining tasks (e.g., node classification, link prediction). Refer to its homepage for more details.