DBLP Article Similarities (DBLP-ArtSim) dataset

This dataset contains similarity scores among articles in AMiner's DBLP v10 dataset. Similarities are calculated using the JoinSim [1] similarity measure on the derived citation network using the following metapaths: Paper - Author - Paper (PAP.csv.gz) Paper - Topic - Paper (PTP.csv.gz) Paper - Venue - Paper (PVP.csv.gz) The Paper to Venue relationships also also provided in PV_relationships.csv.gz. The file aminer_ids.csv.gz contains a mapping from AMiner's ids to our internal numeric ids used in the similarities files. [1] Xiong, Y., Zhu, Y., Yu, P.S.: Top-k similarity join in heterogeneous information networks. IEEE Transactions on Knowledge and Data Engineering 27(6), 1710– 1723 (2015)