Fine-grained Sentiment Analysis Technology Application In Review Mining System

With the continuous development of Internet and e-commerce,people are increasingly keen on online shopping,which leads to a sharp increase in product reviews.The product reviews contain a lot of useful information,consumers can find out the product's reputation by the reviews to make purchasing decision.Branding company can discover the advantages and shortcomings of the product by the reviews,and it can publicize the advantages and improve shortcomings to maintain the brand value better.But the product's reviews on the E-commerce website are divided into two categories of praise and negative,and the traditional emotion analysis could not be applied more finely to the feature attribute,the overall emotional tendencies have been unable to meet the consumer and branding company needs.They want to understand more about the user's specific feature of the product from the comments.Nowadays,there are not many systems use fine-grained sentiment analysis technologies to analysis of reviews and maintain the brand value of product.To solve the problems,this paper focuses on the fine-grained sentiment analysis technologies,and the results of the analysis are applied to the review mining system,it can help consumers select products and help branding companies maintain the brand value of product.The main research contents of this paper are as follows:1.This paper studies text tendency analysis technologies,including text preprocessing,feature extraction and so on,this paper studies and contrasts the emotion analysis technology based on semantic analysis and machine learning.2.Based on the previous feature extraction and classification,this paper presents a method of extracting explicit and implicit features separately.In the process of explicit feature attributes mining,this paper adds three rules to improve the accuracy of feature extraction based on the principle of proximity.In the process of implicit feature attributes mining,this paper presents a bi-iterative method which based on relation between feature words and emotional words to expand feature words and emotional words.3.This paper applies the emotional words which expanded by bi-iterative method to build and expand field lexicon,combining with other lexicons,this paper presents improved method based on the common lexicon-based polarity judgment.It uses the part of speech and linguistic knowledge to quantify the polarity of emotional words,and it can be more effective to compare and analyze the feature s of products.4.This paper designs and realizes the review mining system by combining with web crawler technology,including product reviews acquisition module,data preprocessing module,product feature extraction module,sentiment analysis module of product reviews.Finally,the system test shows that fine-grained sentiment analysis technologies can effectively help consumers and companies to make decisions,the system has strong practical value.