基于神经网络的网络舆情情感分析

Network Public Opinion Sentiment Analysis Based on Neural Network

伴随着互联网的普及和网民的增多,越来越多的网民用户在社交媒体上抒发情感和发表看法。海量的舆情文本产生的同时,对舆论信息在网络上的传播也起到了促进作用。由于中文的舆情信息具有数据信息量较大、文本格式的不规范性以及语义复杂且多元化、语法特殊性、表述隐喻等诸多特点,为了更好地从文本评论信息中分析出用户的情感倾向性观点,对文本数据进行情感分类并挖掘人们对某一事物、某一现象的意见看法和情感态度,文本情感分析技术顺势而生。为了解决上述问题,同时随着深度学习和深层神经网络的广泛流行和研究深入,深度学习算法开始被广泛应用于舆情分析领域的研究中。因此本文对比分析了基于深度学习神经网络的舆情情感分析方法,为多模态情感分析研究工作提供理论依据。相较于传统的情感分析方法,深度学习算法能够学习出海量数据的特征,不再需要人工进行特征的提取和构建,充分对海量文本数据的信息进行自动化学习挖掘,节省了大量的时间和人力物力的同时又具有较强的表达能力,在各种情感分析任务中都具有不错的效果。

With the popularization of the internet and the increase of netizens,more and more netizens express their feelings and opinions on social media.The mass of public opinion text producted at the same time,the dissemination of public opinion information on the network also played a role in promoting.Because of Chi-nese public opinion information with the data large amount of information,the text format is not normative and complex and diverse,grammar,semantics particularity,expression of metaphor,and many other charac-teristics,in order to better information from text comments in the user’s emotion tendentiousness opinion,emotion classification and text data mining for something,a phenomenon of opinion and emotion attitude,text sentiment analysis technology has been born.In order to solve the above problems,with the widespread popu-larity and in-depth research of deep learning and deep neural network,deep learning algorithm has been widely applied in the research of public opinion analysis.Therefore,this paper makes a comparative analysis of public opinion sentiment analysis methods based on deep learning nerual network to provide theoretical ba-sis for multi-modal sentiment analysis research.Compared to the traditional sentiment analysis method,deep learning algorithm can learn the characteristics of huge amounts of data,and no longer need to feature extrac-tion and artificial building,fully automatic learning of massive text data information mining,save a lot of time and manpower and has strong ability to express in all kinds of emotional analysis task has a good effect.