车企舆情正负面情感识别与预测

Emotion Recognition and Prediction of Public Opinions about Automobile Enterprises

为帮助企业获得更多的信息,进一步了解客户,预测和增强客户体验,合理改进产品性能,通过情感词典来对汽车行业的网络舆情进行分析与预测.首先对预处理后的文本进行分词,提取关键词,绘制词云图,初步判定舆情中人们关注的热点.然后利用训练集数据对情感词典进行训练,提取文本特征,并采用基于情感词典的传统情感分类法进行文本情感识别分类.分类结果显示,训练集的预测准确率为85.73%,测试集的准确率为83.62%.最后利用LDA模型对文本进行主题分析,得到正面、负面文本数据的第一主题与第二主题.

With the development of the Internet,there are a large number of online public opinions in various industries.In this paper,positive and negative emotion identification and prediction are carried out according to the public opinion of automobile enterprises.First of all,word segmentation,keyword extraction and word cloud map are carried out on the pre-processed text to preliminarily determine the hot spots that people pay attention to in the public opinion.Then,the Sentiment Dictionary is trained to extract text features with the data from the training set,and the traditional Sentiment Classification based on Sentiment Dictionary is used for text sentiment recognition and classification.The classification results showed that the prediction accuracy of the training set was 85.73%,while the accuracy of the test set was 83.62%.Finally,LDA model is used to analyze the topic of text,and the first topic and second topic of positive and negative text data are obtained.