Identifying Human Phenotype Terms by Combining Machine Learning and Validation Rules

Table 2

The performance of the different types of used features for IHP: linguistic (L), orthographic (O), morphological (M), context (C), lexical (LE), and others (X). These features were tested in a single cross-validation iteration.

	Precision	Recall	-measure
Baseline	0.452	0.594	0.514
L	0.463	0.72	0.564
O	0.452	0.594	0.514
M	0.457	0.766	0.573
C	0.469	0.783	0.587
Le	0.453	0.606	0.518
X	0.428	0.697	0.530
L + O	0.458	0.720	0.560
L + O + M	0.451	0.760	0.566
L + O + M + C	0.478	0.800	0.598
L + O + M + C + Le	0.478	0.805	0.600
L + O + M + C + Le + X	0.482	0.823	0.608