Skip to main content

Table 5 Comparison with other systems on AIMED corpus

From: Learning an enriched representation from unlabeled data for protein-protein interaction extraction

Method or author F-score AUC
(Miwa et al., 2009) [16] 65.2 89.3
Our method (Combined) 63.5 87.2
(Miwa et al., 2009) [12] 62.7 (64.3) 83.2 (87.9)
Our method (Lex) 61.4 86.11
Our method (FCD) 60.1 83.8
(Miyao et al., 2008) [11] 59.5 -
(Airola et al., 2008) [10] 56.4 84.8
(Sætre et al., 2007) [9] 52.0 -
(Mitsumori et al., 2006) [14] 47.7 -
  1. For the work [12], the scores in brackets are their reported results obtained by removing all the examples with self-interaction protein pairs in AIMED corpus.