Speaker intent detection and semantic slot filling are two critical tasks inspoken language understanding (SLU) for dialogue systems. In this paper, wedescribe a recurrent neural network (RNN) model that jointly performs intentdetection, slot filling, and language modeling. The neural network model keepsupdating the intent estimation as word in the transcribed utterance arrives anduses it as contextual features in the joint model. Evaluation of the languagemodel and online SLU model is made on the ATIS benchmarking data set. Onlanguage modeling task, our joint model achieves 11.8% relative reduction onperplexity comparing to the independent training language model. On SLU tasks,our joint model outperforms the independent task training model by 22.3% onintent detection error rate, with slight degradation on slot filling F1 score.The joint model also shows advantageous performance in the realistic ASRsettings with noisy speech input.
展开▼
