Researcher data for pre-training

This data was extracted from the DBLP, ACM, and MAG digital library as the input of the researcher pre-training model RPT[1][2].This dataset contains a researcher semantic document set and a researcher community graph.--author_document_corpus.txtthe researcher semantic document set, where each line represents a researcher's semantic document set, including the researcher ID and document sequence separated by tab(\t), for example:a1 \t d1 \t d2 \t ... \t dm \na2 \t d1 \t d2 \t ... \t dn \n...--author_community.pklwhere the key is the researcher id, and the value includes two vectors of equal length, preserving the neighbor id and corresponding relation type index.

For example:{a1: ["neighbors": [a2, a4, a9, ..., a6], "relations": [0, 0, 1, ..., 2] ] a2: ...} [1] Qiao, Ziyue, et al. "RPT: Toward Transferable Model on Heterogeneous Researcher Data via Pre-Training." arXiv preprint arXiv:2110.07336 (2021).[2] github.com/joe817/RPT