Slot Filling Using En-Training

The performance of slot filling is crucial for spoken language comprehension. Aiming at the problem of low filling accuracy, an En-training model for slot filling is proposed based on the idea of ensemble learning. The structure completes the task of slot filling by constructing and combining multiple classifiers. Experiments are carried out on ATIS data sets, the results show that the En-training adaptive network structure proposed in this paper has a significant improvement in accuracy, recall rate and F1 value.