Joint Multiple Intent Detection and Slot Filling Via Self-Distillation

This paper addresses zero-shot slot filling, which tries to build a system that can generalize to unseen slot types without any training data

Speech Intent Classification and Slot Filling Different from its counterpart in Natural Language Understanding that takes text as input, here the

filling slot The goal of Slot Filling is to identify from a running dialog different slots, which correspond to different parameters of the user's query

casino best slot games The existing works either treat slot filling and in- tent detection separately in a pipeline man- ner, or adopt joint models which sequentially label slots