融合主题模型的图神经网络对话情感识别  

Fusion of Topic Models for Conversational Emotion Recognition in Graph Neural Networks

作　　者：张甜甜 李众 谷一宽 杨晓霞 ZHANG Tiantian;LI Zhong;GU Yikuan;YANG Xiaoxia(School of Software,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学软件学院,山西太原030051

出　　处：《中北大学学报（自然科学版）》2024年第3期286-295,共10页Journal of North University of China(Natural Science Edition)

基　　金：山西省自然科学基金资助项目(20210302123019)。

摘　　要：对话情感识别(ERC)旨在预测对话中语句的情感类别。目前,基于图神经网络的ERC方法主要采用固定的超参数来确定图中边的连接,缺乏针对不同数据进行自适应构边的策略,且忽略了语句间的主题关系。此外,在图神经网络的训练过程中,这些方法通常采用求和叠加的方式来聚合节点信息,限制了模型的非线性能力。为此,本文将主题模型与图神经网络相融合,提出了一种新的构边方法。首先利用主题模型获取对话中语句的主题分布,然后将具有相同主题的语句相互连接。同时,引入了SwiGLU门控单元,用于调控图神经网络中层与层之间的信息流动。在边的类型方面,考虑了人物信息的差异,以更好地捕捉情感变化的内因和外因。通过在4个公开数据集(IEMOCAP、MELD、EmoryNLP、DailyDialogue)上进行的广泛实验,与当前先进的ERC方法相比,本文的方法在前3个数据集上的F1分数分别提升了1.69%,0.27%和0.38%。此外,本文的自适应方法在长对话上的效果提升了2.11%,优于短对话的0.8%,同时,通过引入SwiGLU有效减缓了图神经网络中的过度平滑现象。综合结果表明,本文提出的融合主题模型进行自适应构边以及引入SwiGLU门控单元的图神经网络方法,能够有效提高对话情感识别的效果,增强模型的鲁棒性。Emotion recognition in conversations(ERC)aims to predict emotional categories of utterances within a conversation.Presently,graph neural network-based ERC methods predominantly employ fixed hyperparameters to determine the connections among graph edges,lacking adaptive strategies for edge con-struction tailored to diverse data and ignoring thematic relationships between statements.Furthermore,during the training process of graph neural networks,these methods often utilize a summation superposi-tion approach to aggregate node information,limiting the model's non-linear capabilities.To address these limitations,this paper integrated topic modeling with graph neural networks and proposed a novel edge construction method.Firstly,a topic model was employed to extract the thematic distribution of state-ments within a conversation,followed by the connection of statements sharing same themes.Meanwhile,the SwiGLU gated unit was introduced to regulate the flow of information between layers in the graph neu-ral network.Considering differences in character information,the edge types were carefully tailored to bet-ter capture intrinsic and extrinsic factors influencing emotional changes.Through extensive experiments conducted on four publicly available datasets(IEMOCAP,MELD,EmoryNLP,DailyDialogue),our approach demonstrates significant improvements over advanced ERC methods,achieving F1 score enhancements of 1.69%,0.27%,and 0.38%on the first three datasets,respectively.Moreover,our adaptive method exhibits a 2.11%improvement on long conversations,surpassing the 0.8%gain on short conversations.The introduction of the SwiGLU unit effectively mitigates over-smoothing phenom-ena in the graph neural network.Consequently,the proposed approach,which combines adaptive graph construction with topic modeling and integrates SwiGLU gated units into graph neural networks,proves to be effective in enhancing dialogue emotion recognition,thereby reinforcing the model’s robustness.

关 键 词：对话情感识别 图神经网络 主题模型 门控单元 图结构 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]