%0 Journal Article
%T 基于SenticNet的电商客服对话细粒度情感波动分析<br>Fine-Grained Emotion Fluctuation Analysis of E-Commerce Customer Service Dialogues Based on SenticNet
%A 朱陈云婧
%A 马彪
%J Service Science and Management
%P 24-36
%@ 2324-7916
%D 2025
%I Hans Publishing
%R 10.12677/ssem.2025.141004
%X 电商的迅速发展为消费者带来了便捷&#65292;但客户服务面临的挑战也日益增多。客服响应效率与问题解决效果对用户体验至关重要&#65292;然而目前广泛采用的事后评价机制无法对用户情绪进行即时监控&#65292;难以及时识别和处理用户情绪波动&#65292;限制了服务满意度的提升。为克服这一不足&#65292;文章提出了一种基于SenticNet情感词辅助的跨句上下文对话情绪分类模型(SLED-EC)。该模型不仅整合了词语级与句子级的情感识别任务&#65292;以更精细地捕捉情感细节&#65292;还通过分析上下文信息来识别情绪的转变点&#65292;构建情感波动曲线&#65292;动态捕获多轮对话中的情感变化。实验结果表明&#65292;SLED-EC在准确率(67.78%)、召回率(83.81%)和F1值(72.36%)方面优于基准模型&#65292;展示了其在客服对话场景中情感分析的优势。该研究为电商客服系统提供了一种新的解决方案&#65292;有助于提升用户满意度和优化服务体验。<br>The rapid advancement of e-commerce has significantly enhanced consumer convenience, yet it has simultaneously exacerbated the challenges faced by customer service. Customer service response efficiency and problem-solving effectiveness are critical to the user experience. However, widely adopted post-feedback mechanisms are inadequate for real-time monitoring of user emotions, hindering timely identification and management of emotional fluctuations, which limits improvements in service satisfaction. To address this shortcoming, this paper proposes a SenticNet lexicon-assisted cross-sentence contextual dialogue emotion classification model (SLED-EC). The proposed model integrates both word-level and sentence-level emotion recognition tasks, enabling more precise capture of emotional nuances. Additionally, it analyzes contextual information to identify emotional transition points and constructs emotional fluctuation curves, dynamically capturing emotional changes in multi-turn dialogues. Experimental results demonstrate that SLED-EC significantly outperforms baseline models in accuracy (67.78%), recall (83.81%), and F1-score (72.36%), underscoring its advantages in emotion analysis for customer service dialogue scenarios. This study offers a novel solution for e-commerce customer service systems, contributing to improved user satisfaction and optimized service experience.
%K 对话情感分析&#65292
%K 上下文依赖&#65292
%K 情绪波动曲线&#65292
%K 电商客服<br>Dialogue Sentiment Analysis
%K Cross-Sentence Context
%K Emotion Fluctuation Curve
%K E-Commerce Customer Service
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=104480