We utilize two categories of datasets in our project: OGBN and Amazon Review.
The Amazon Review dataset is available from this link. This dataset is compiled based on Yan's research work. We re-split it for few-shot node classification. The label files are located under the path split/class_split.json for each dataset.
The OGBN dataset is set to be downloaded automatically. However, users are required to follow the instructions provided on the OGBN official website to obtain the raw text for the 'ogbn-arxiv' and 'ogbn-products' datasets. We also provide the corresponding Baidu Cloud link.
To ensure the reproducibility of our results, please copy the corresponding files from the 'preprocess/' folder into the 'ogbn_arxiv' and 'ogbn_products' folders. This step is crucial as it guarantees the same data split is used.