Skip to main content

Table 8 Performance of our current system in the GE task in contrast to TEES, the best case (B-C) and worst case (W-C) oracles with re-ranked output, as well as the worst case oracle with randomized rankings (averaged over ten runs).

From: Application of the EVEX resource to event extraction and network construction: Shared Task entry and result analysis

All events P R F
B-C oracle (re-ranked) 81.32 39.61 53.27
W-C oracle (re-ranked) 54.92 39.61 46.02
W-C oracle (random) 51.06 39.19 44.34
Current system 47.15 39.61 43.05
TEES 45.46 40.39 42.77
Single-arg. events    
B-C oracle (re-ranked) 81.37 50.58 62.38
W-C oracle (re-ranked) 56.09 50.58 53.19
W-C oracle (random) 52.73 50.00 51.33
Current system 48.66 50.44 49.53
TEES 47.16 51.09 49.04
Multiple-arg. events    
B-C oracle (re-ranked) 81.02 16.83 27.87
W-C oracle (re-ranked) 48.61 16.83 25.00
W-C oracle (random) 42.66 16.75 24.05
Current system 39.64 17.12 23.91
TEES 37.57 18.17 24.50