基于SVM的淘宝商品评论情感分析

北京理工大学珠海学院2020届本科生毕业论文Analysis of emotional in Taobao product reviews based on SVMAbstractThe popularity of e-commerce has greatly promoted the development of online shopping.With the development of cloud technology,more and more data are generated,resultingin a lot of comment information involving goods and online shopping services,which areparticipated by consumers.After each consumption,online shopping users will leave cor-responding data,including the evaluation of product quality,price,business service,etc.Using natural language processing technology to quantify these data,using emotional anal-ysis,it can be transformed into valuable messages.This is of great value to both consumersand businesses.Therefore,the data in this paper is e-commerce review data,which is characterized by Chi-nese as the main language.Therefore,the working principle of this paper is mainly toextract the entities of sentences from Taobao comments through the Chinese word segmen-tation system,that is,the parts with clear emotional tendency in the sentences,so as to judgethe attitude of consumers towards the goods.In the process of text processing,it involvesnatural language processing technology,such as text feature extraction by quantifying nat-ural language;in addition,as a supervised learning,support vector machine also involvesthe field of machine learning,so this paper will also evaluate the excellence of the model incombination with other al gorithms in the end.In this paper,4000 Taobao reviews are analyzed and divided into training set and test setin the proportion of 8:2.First of all,before fitting the model,the data were observed andanalyzed,and the dimension was unified by using the standardization part,and word seg-mentation was carried out.Secondly,try to train the model with SVM.After getting a goodfit,the performance and prediction ability of the model are tested.Finally,through the goodmodel after the training of the data,the relevant improvements are proposed.Keywords:Sentiment analysis Chinese word segmentation support vector machinee-commerce review data