. 2012 Jan 30;5(Suppl 1):137–145. doi: 10.4137/BII.S8963

Table 2.

Performance of the SVM classifier with different feature combinations on the testing data.

	Feature set	Micro-averaged F-measure	Precision	Recall
N-gram feature	N_u	0.4492	0.5971	0.3601
	N_u + N_b	0.4707	0.6505	0.3687
	N_u + N_b + N_t	0.4542	0.6128	0.3609
Knowledge-based features	N_u + N_b + K_m	0.4623	0.5946	0.3781
	N_u + N_b + K_l	0.4650	0.6161	0.3734
	N_u + N_b + K_m + K_l	0.4750	0.6525	0.3734
Syntactic features	N_u + N_b + K_m + K_l + S_d	0.4781	0.6667	0.3726
	N_u + N_b + K_m + K_l + S_p	0.4783	0.6553	0.3766
	N_u + N_b + K_m + K_l + S_t	0.4798	0.6584	0.3774
	N_u + N_b + K_m + K_l + S_t + S_p	0.4818	0.6612	0.3789
	N_u + N_b + K_m + K_l + S_t + S_p + S_d	0.4804	0.6657	0.3758
Context features	N_u + N_b + K_m + K_l + S_t + S_p + C_m	0.4697	0.6218	0.3774
	N_u + N_b + K_m + K_l + S_t + S_p + C_p	0.4758	0.6508	0.3750
	N_u + N_b + K_m + K_l + S_t + S_p + C_m + C_p	0.4787	0.6593	0.3758
Class-specific features	N_u + N_b + K_m + K_l + S_t + S_p + R	0.4883	0.6667	0.3852
Class-specific features	N_u + N_b + K_m + K_l + S_t + S_p + R + F	0.4878	0.6694	0.3837
All	All features	0.4720	0.6279	0.3781