“…This figure was taken from the original publication. 2 (A) The distribution of the F1 value for each named entity, which aggregates the results of five different named-entity-recognition methods, namely Lattice, 4 TENER, 5 GraphNER, 6 FLAT, 7 and our CCIE (denoted as “★”). The colored distributions correspond to different named entities; (B) the distribution of the F1 value for each text category, which aggregates the results of eight text classification methods, namely Transformer, 9 DPCNN, 10 FastText, 11 TextCNN, 12 TextRNN, 13 TextRCNN, 14 LSTM, 15 and our CCIE.…”