Research Article

Semi-Supervised Learning of Statistical Models for Natural Language Understanding

Table 3

Performance comparison between the proposed framework and three other approaches (HF denotes the hybrid framework and DT denotes discriminative training the HVS model.

Measurement
HVS
HF
DT
Proposed framework
CRFs HM-SVMs

Recall (%) 87.81 90.99 91.49 92.08 92.04
Precision (%) 88.13 90.25 91.87 93.83 94.36
-measure (%) 87.97 90.62 91.68 92.95 93.18