Skip to main content
. 2005;2005:420–424.

Table I.

Semantic Type Distribution in Training and Testing Data Sets

Category Training (n=1281) Testing (n=293)
Clinical 33.4% 20.8%
Technical 14.6% 18.1%
Backchannel 27.2% 34.5%
Miscellaneous 24.7% 26.6%