Table 8.

The comparison of the dialog state tracking F1-scores between the proposed model and DSTC4 participants.

Model	Precision	Recall	F1-score
String match [26]	0.3589	0.1925	0.2506
LSTM [20]	0.3400	0.2010	0.2530
Proposed model	0.5173	0.3499	0.4174