Research And Application Of Sentiment Analysis Based On Products Reviews

With the rapid development of e-commerce,the rapid development of various ecommerce platforms is also changing consumers' shopping methods and shopping experience,and has also greatly changed our lifestyle.Among them,the data that most intuitively reflects the user's view of the product and the e-commerce platform is the product's review data.How to find a representative topic and emotional tag in a large number of review data sets is one of the research focuses of sentiment analysis.These data can not only reflect the user's opinion on the product,but also can extract the user's emotional information,to provide commercial reference value for more users and ecommerce platform,product recommendation,product improvement and the same products provide a way to compare with each other.Therefore,sentiment analysis based on product review data has become the focus of businesses and users,and it is also an important content of this research.The main contents of this paper are as follows:(1)This paper studies the LDA theme model,and based on this,proposes a LDA emotional theme model based on reviews.The model analyzes the problem of short text emotional topics,proposes an unsupervised algorithm for analyzing emotional topics in short texts by combining LDA topic models with emotional factors,and builds an emotional topic model based on this algorithm.(2)This paper designs an algorithm based on emotion dictionary to calculate the value of word emotion tendency,and improves the traditional SO-HowNet algorithm.The traditional sentiment orientation algorithm only calculates the emotional value of a single word.It lacks practical significance and does not consider the weight of words in the document.In this paper,for the problem of sentiment orientation,the weights and general importance of TF-IDF metrics are introduced.At the same time,the traditional sentimental value algorithm is used to calculate the value of emotional sentiment.(3)This paper designs an emotional theme analysis model based on commodity reviews.The model builds a computable Chinese word vector dictionary by using Chinese word segmentation and the algorithm for constructing word vectors,and then uses the LDA emotional theme model based on reviews to analyze the emotional subject related word bags.Combining the algorithms based on the sentiment lexicon to calculate the sentimental value of the word,the positive and negative evaluations are obtained.Through experimental tests,the results show that the proposed model has a certain degree of improvement in accuracy and recall.(4)Finally,this paper designed the emotional theme analysis system,and completed the design of the overall function of the system.Through the data input module,data analysis module,data display module structure and function of detailed design,through the data visualization to achieve the emotional theme analysis system to help users analyze the emotional data in the comments data.