摘要
针对传统就业情况问卷调查方法无法分析诸多变量间的复杂关系且严重依赖人工的问题,建立了一种基于集成深度学习ATE与APC的联合学习模型LCF-ATEPC。该模型集成了文本情感分析中局部上下文聚焦及BERT机制,同时通过文本方面项提取和情感极性分类两个子任务的交互,克服了常规模型中方面项提取任务精度不足的问题。数值实验结果表明,通过LCF-ATEPC算法挖掘社交媒体数据,并进行多方面话题情感分析,可有效提升分析结果的准确度,且相较传统人工评估与深度学习算法分别提升了约3.58%和1.12%,既提高了工作效率又降低了人力成本。
In view of the problem that the traditional questionnaire survey method can't analyze the complex relationship between many variables,and it is heavily dependent on labor.This paper establishes a joint learning model LCF-ATEPC based on integrated deep learning ATE and APC.The model integrates local context focusing and BERT mechanism in text emotion analysis.Through the interaction of two subtasks of text aspect item extraction and emotion polarity classification,the problem of insufficient accuracy of aspect item extraction task in conventional model is overcome.The numerical experimental results show that the accuracy of the analysis results can be effectively improved by mining social media data and conducting various topic emotion analysis through LCF-ATEPC algorithm,which is about 3.58% and 1.12% higher than the traditional manual evaluation and deep learning algorithms,respectively.It not only improves the work efficiency,but also reduces the labor cost.
作者
高晓梅
张永红
GAO Xiaomei;ZHANG Yonghong(Xi’an Aeronautical Polytechnic Institute,Xi’an 710089,China)
出处
《电子设计工程》
2023年第21期51-55,共5页
Electronic Design Engineering
基金
陕西省教育科学十三五规划2020年度课题(SGH20Y1657)。
关键词
集成深度学习
大数据分析
社交媒体挖掘
就业质量评估
多方面情感分析
integrated deep learning
big data analysis
social media mining
employment quality ass-essment
multifaceted emotion analysis