A Novel Feature Selection Strategy for Enhanced Biomedical Event Extraction Using the Turku System

Table 7

The best feature combination after choosing features dynamically in trigger and edge detection.

	1	2	3	4	5	6
Feature in trigger	#Sentence feature	#Main feature	#Linear order	#Content feature	#Attached edge	#Chain feature
			Feature		Feature
Feature size	18998	24944	73744	8573	100561	178345
(Merely one feature analysis)
𝐹-score under one class	0	42.05	3.50	27.11	7.33	5.48
Average contribution	0	0.001	4𝑒−5	0.003	7𝑒−5	3𝑒−5
(Double feature combination analysis)
Best performance	(+4)	(+4)	(+4)	(+1)	(+2)	(+4)
Involving ith feature	45.39	36.65	22.29	45.39	28.27	23.48
Worst performance	(+2)	(+1)	(+1)	(+3)	(+1)	(+1)
Involving ith feature	0	0	0	22.29	0.91	3.10
(Three-feature combination analysis)
Best performance	(+4, 5)	(+1, 4)	(+1, 4)	(+1, 5)	(+1, 4)	(+1, 4)
Involving ith feature	49.44	47.82	45.13	49.44	49.44	45.51
Worst performance	(+2, 3)	(+1, 3)	(+1, 2)	(+3, 6)	(+1, 3)	(+1, 3)
Involving ith feature	0.11	0.11	0.11	20.76	1.98	2.40
(Four-feature combination analysis)
Best performance	(+2, 4, 5)	(+1, 4, 5)	(+1, 4, 5)	(+1, 2, 5)	(+1, 2, 4)	(+1, 2, 4)
Involving ith feature	50.71	50.71	49.74	50.71	50.71	48.82
Worst performance	(+2, 3, 6)	(+1, 3, 6)	(+1, 2, 6)	(+3, 5, 6)	(+1, 2, 3)	(+1, 2, 3)
Involving ith feature	5.77	5.77	5.77	21.13	7.02	5.77
(Five-feature combination analysis)
Performance
Without ith feature	34.22	47.1	49.90	16.37	50.19	51.34