摘要
随着深度学习技术在自然语言处理领域的广泛应用,短文本情感分类技术得到显著发展。该文提出了一种融合TextCNN-BiGRU的多因子权重文本情感分类算法。算法通过引入词语情感类别分布、情感倾向以及情感强度三个关键因子改进了词语的向量表示。将基于词向量表示的短文本分别作为TextCNN和BiGRU模型的输入,提取文本关键局部特征以及文本上下文的全局特征,将两种特征进行线性融合,实现中文短文本的情感分类。在公开的两个情感分类数据集上验证了多因子权重向量表示方法和融合TextCNN-BiGRU的情感分类模型的有效性,实验结果表明,文中提出的算法较单一模型在短文本情感分类准确率上提高了2%。
With the widespread application of deep learning technology in natural language processing,the research on sentiment classification of short texts has seen significant advancements.A multifactorial weighted text sentiment classification algorithm that integrated TextCNN and BiGRU is proposed.The algorithm enhanced Word2Vector representations by introducing distribution of word sentiment category,sentiment tendency,and sentiment intensity.Subsequently,short texts based on word vector repres⁃entations were input into the TextCNN and BiGRU models,extracting local features of the text and the global context features.These two sets of features were linearly fused to achieve sentiment classification of Chinese short texts.This paper validated the effectiveness of the integrated TextCNN-BiGRU sentiment classification model on two public sentiment classification datasets.The experimental results show that the proposed algorithm improves the accuracy of short text emotion classification by 2%compared with the single model.
作者
鲁富宇
冷泳林
崔洪霞
LU Fuyu;LENG Yonglin;CUI Hongxia(College of Information Science and Technology,Bohai University,Jinzhou 121003,China)
出处
《电子设计工程》
2024年第10期44-48,53,共6页
Electronic Design Engineering
基金
国家自然科学基金资助项目(62006035)
辽宁省教育厅重点攻关项目(JYTZD2023174)。
作者简介
鲁富宇(1980-),男,辽宁阜新人,硕士,讲师。研究方向:短文本情感分析和知识图谱存储。
