基于深度学习的意图识别与语义槽填充联合建模研究

Joint Modeling of Intention Recognition and Semantic Slot Filling Based on Deep Learning

意图识别和槽填充是自然语言理解的两个基本任务,它们之间互相携带了对方的信息。本文提出了一种基于BERT-CRF的联合识别模型。实验结果表明,该模型在意图识别的准确率和语义槽填充的F1分数方面都有显著的提高。

Intention recognition and slot filling are two basic tasks of natural language understanding.Its carry each other's information.This paper presents a joint recognition model based on BERT-CRF.Experimental results show that the proposed model can improve the accuracy of intention recognition and the F1 score of semantic slot filling.