Table 4.

The performance of Random Forest classifiers for each category tested on MayoConnect (MC) and Cancer Survivor’s Network (CSN) data and trained on either MC or CSN data, using local context features.

Category	Test MC data						Test CSN data
	Train MC data			Train CSN data			Train CSN data			Train MC data
	Preca	Recall	F-measure	Prec	Recall	F-measure	Prec	Recall	F-measure	Prec	Recall	F-measure
Medical	.90	.91	.90	.71	.71	.71	.78	.75	.76	.71	.67	.68
Social	.85	.85	.85	.61	.69	.66	.83	.82	.82	.75	.77	.76
Psychological	.77	.76	.76	.51	.55	.51	.73	.74	.73	.50	.54	.52
Background	.77	.77	.77	.51	.53	.51	.74	.74	.74	.50	.52	.51
Wellness	.80	.79	.79	.55	.59	.56	.70	.71	.70	.51	.60	.55
Physical	.82	.83	.83	.56	.54	.55	.70	.70	.70	.60	.66	.63
Previous	.58	.58	.58	.54	.60	.57	.71	.71	.71	.55	.58	.56
Other	.84	.86	.85	.65	.56	.60	.67	.66	.66	.61	.63	.62

^aPrec: precision.