Benchmark Dataset for Impact4Cast: Forecasting high-impact research topics via machine learning on evolving knowledge graphs

This dataset, from the paper 'Forecasting high-impact research topics via machine learning on evolving knowledge graphs' by Xuemei Gu and Mario Krenn, includes benchmark data for evaluating fully connected NNs, Transformers, Random Forest, and XGBoost on prediction tasks with 2-4 year training and 1-5 year evaluation intervals across two impact ranges (IR). It also provides some examples about 10M evaluation samples (2019-2022) with raw outputs from a neural network trained on 2016-2019 data.