Table 5.

Classification accuracy comparisons (in terms of precision, recall and F₁ scores) for the strongly aggravating and alleviating classes of interactions found in E-MAPs.

Dataset	Method	Alleviating			Aggravating
		Precision	Recall	F1	Precision	Recall	F1
Chromosome	wNN	0.66	0.14	0.23	0.71	0.40	0.51
	LLS	0.65	0.07	0.13	0.74	0.38	0.50
RNA	wNN	0.69	0.14	0.23	0.72	0.39	0.51
	LLS	0.75	0.11	0.19	0.72	0.35	0.47
Pombe	wNN	0.64	0.17	0.27	0.70	0.49	0.58
	LLS	0.74	0.09	0.16	0.69	0.50	0.58
Signalling	wNN	0.71	0.06	0.11	0.65	0.27	0.38
	LLS	0.50	0.01	0.02	0.65	0.32	0.43
ESP	wNN	0.78	0.14	0.24	0.64	0.42	0.51
	LLS	0.67	0.09	0.16	0.66	0.42	0.52

The highest value for each dataset is highlighted in bold. While the neutral class has been left out for the sake of clarity, note that in all cases recall was ≈ 0.99 and precision was > 0.96.