A neural framework for joint prediction on intent identification and slot filling

Abstract

In task-oriented dialog systems, understanding of users’ queries (expressed in natural language) is a process of parsing users’ queries and converting them into some structure that machine can handle. The understanding usually consists of two parts, namely intent identification and slot filling. To address this problem, we propose a neural framework, named SI-LSTM, that combines two tasks and integrates CRF into LSTM network, where the slot information is extracted by using CRF, and the intent will be identified by using LSTM. In our approach, the slot information is used for determining the intent, while the intent type is used to rectify the slot filling deviation. Based on the dataset provided by NLPCC 2018, SI-LSTM achieved 90.71% on intent identification, slot filling and error correction in terms of accuracy.