基于降噪自动编码器及其改进模型的微博情感分析

发布时间：2018-04-05 23:31

本文选题：降噪自动编码器　切入点：微博　出处：《计算机应用研究》2017年02期

【摘要】：随着自然语言处理科学的迅猛发展,情感分析作为其重要的一个分支广泛应用于社交网络平台上,尤其是微博由于其传播广泛且蕴涵丰富的情感信息而备受学者青睐。为解析微博中表达的情感信息以及深入挖掘其蕴涵的潜在感情,在降噪自动编码器的深度模型之上研究探索改进了这个深度学习模型。降噪自动编码器的工作特点是在引入噪声的干扰之下实现对原始输入的还原,而其改进模型的优势在于考虑到了噪声的多样性和复杂性,并通过深度学习训练加强模型的原始特征复原能力,以此来克服不可预判的原始输入噪声。通过分别使用SVM、降噪自动编码器模型以及改进的模型进行情感分析实验,对比分类效果而得出改进的深度模型对微博文字情感把握更准确而且抗干扰能力及鲁棒性有所提升的结论。
[Abstract]:With the rapid development of natural language processing science, emotion analysis, as an important branch of natural language processing, is widely used in social network platform, especially Weibo is favored by scholars because of its wide spread and rich emotional information.In order to analyze the emotional information expressed in Weibo and dig out the potential emotion, the depth learning model is improved based on the depth model of the noise reduction automatic encoder.The working characteristic of the noise reduction automatic encoder is to restore the original input under the interference of the noise. The advantage of the improved model is that the diversity and complexity of the noise are taken into account.In order to overcome the unpredictably original input noise, the restoration ability of the original features of the model is enhanced through in-depth learning training.By using SVM, the noise reduction automatic encoder model and the improved model to carry on the emotion analysis experiment,By comparing the classification effect, the conclusion that the improved depth model is more accurate and the anti-interference ability and robustness is improved.
【作者单位】：南京工业大学电气工程与控制科学学院;
【分类号】：TP391.1

【相似文献】