TL;DR通过深度学习模型,本研究提出了一种基于度量学习和 Siamese 网络架构的方法来进行情感识别对话任务,取得了 57.71 的宏 F1 得分,相较于相关研究有了显著的提升。
Abstract
The advent of deep learning models has made a considerable contribution to
the achievement of emotion recognition in conversation (ERC). However, this
task still remains an important challenge due to the pluralit