Skip to main content

Table 4 Average test accuracies and F1 for multi-class relation extraction with all instances

From: Incorporating representation learning and multihead attention to improve biomedical cross-sentence n-ary relation extraction

Method Multi-class
  Ternary accuracy Ternary F1 Binary accuracy Binary F1
GS GLSTM [12] 82.3 76.1 82.1 75.8
GCN (K=0) [15] 78.1 74.6 73.1 70.2
AGGCN [16] 79.7 75.5 77.5 73.5
Multihead attention 86.8 84.3 91.6 88.8
With KG 89.8* 86.8* - -