Analysis Of Emotional Tendency Of Commerce Comments Text Based On SVM

In today's world,with the rapid development of e-commerce platform,a large number of user comment texts are produced.These commentary texts contain buyers' emotional tendencies towards products,such as "positive" and "negative".Studying the emotional tendencies of these commentary texts will help platform managers to understand the advantages and disadvantages of products and facilitate future product improvement.At the same time,after obtaining the emotional information in the comment text,potential users are more likely to make correct consumption decisions.The research direction of this paper is to analyze the emotional orientation of comment text on e-commerce platform.It is found that the quality of vector space constructed by the model is not good enough in the process of experiment and research.In the data set of long comment text and short comment text,the performance of SVM and MNB algorithm is just the opposite,so the model constructed by the two algorithms alone does not have general applicability.In order to solve these problems,two improved methods are proposed in this paper.Firstly,corpus and affective dictionary are put into Word2 vec model to construct two word vector spaces and combine them with features.Secondly,traditional machine learning algorithm SVM is improved to combine it with MNB algorithm to construct classifier.The main content of this paper consists of the following parts:Firstly,this paper introduces the current research status of text sentiment orientation analysis,gives a brief overview of the current mainstream methods of text sentiment orientation analysis,analyses the current situation and shortcomings of existing methods,and introduces the commonly used methods of text vectorization and text sentiment classification.Secondly,the Word2 vec algorithm is deeply studied,and a feature optimization algorithm based on Word2 vec is proposed.The algorithm adds large corpus and affective dictionary,combines common features and specific features,and combines and optimizes features.Thirdly,this paper briefly introduces the SVM and MNB algorithms,and deeply studies their respective advantages in affective classifier.On this basis,SVM algorithm and MNB algorithm are fused to construct SVM-MNB emotional classification model.Finally,this paper validates the improved model through relevant experiments,and compares and analyses it with other models.