The MAPLE Benchmark for Graph Mining

This repository contains the graph format of MAPLE (a large-scale collection of scientific papers across 19 scientific fields), which can serve as a comprehensive evaluation benchmark for graph mining tasks (e.g., node classification, link prediction) in the scientific domain.

MAPLE was introduced in the WWW 2023 paper "The Effect of Metadata on Scientific Literature Tagging: A Cross-Field Cross-Model Study", available at https://arxiv.org/abs/2302.03341.

The original format of MAPLE used for text mining tasks (e.g., multi-label text classification) can be found at https://zenodo.org/record/7611544.

Please refer to https://github.com/yuzhimanhua/MAPLE for more details on the data format.

If you find MAPLE useful, please cite our paper:

@inproceedings{zhang2023effect,
title={The effect of metadata on scientific literature tagging: A cross-field cross-model study},
author={Zhang, Yu and Jin, Bowen and Zhu, Qi and Meng, Yu and Han, Jiawei},
booktitle={WWW'23},
year={2023}
}