The Analysis And Application Of User Comments Base On E-commemerce

With the rapid development of e-commerce,more and more users make online shopping through the e-commerce platform and leave a large number of user comments on the platform.These user comment data contain a lot of user experience,which is of great data value for enterprises to understand users' voice and quickly discover users' concerns and needs.Therefore,this paper takes the children's phone watch industry of jingdong platform as an example to study and analyze the application value of user comments to enterprises through three links: data collection and cleaning,extraction of key words and phrases and emotion analysis,analysis and application of comment data.First,through the python browser from jingdong electric business platform for children phone watch industry users comments and its related content,first of all,this article through the python simulation browser from jingdong electric business platform for children phone watch the user comments and relevant content of the industry,but the data collected by the electric business platform is not normative,so clean to brand attribute data and user comments separately.How to apply the user comment data after cleaning is the focus of this paper,and it is also an urgent problem for enterprises to solve.By building a commodity index system,this paper indexes user comments,and quickly analyzes and excavates users' views on commodities and services.The construction of the commodity index system includes three parts,namely,the design of commodity index,the extraction of index keywords,the extraction of keyword phrases and emotional classification.Word segmentation based on user comments,and according to user's mention of word frequency and the commodity itself attribute parameter form commodity index,at the same time through the TF-IDF keyword extraction algorithm are keyword extraction from user reviews,combination index keywords that regular expression from the user comments extract keyword phrase,finally through to the classification model,Sklearn Snownlp under various classification algorithms and the LSTM under keras,CNN model respectively for keyword phrase emotional classification and evaluation,select the keyword phrase emotion of CNN for optimal classification model.At the same time,in the process of extracting index keywords,in order to solve the problem of diversified language expression,vectorized the words commented by users through Word2 vec,and extracted the words with high similarity to the word vectors of index keywords,so as to realize the expansion of index keywords.After data cleaning and indicator system building,indexed user comments are formed.We form different data application modules through data analysis methods and dimensions,and analyze them with actual cases.we understand the overall situation of the children's phone watch industry by describing the statistical analysis method,including the leading brands,and whether there are opportunities in the market.At the same time,through the comparison between brands and commodity indicators,it can be known that users are most concerned about the battery capacity and battery life of children's phone watches,and different brands in the same indicators between the performance of some differences.At the same time,according to the analysis results,the enterprise can give priority to the improvement and optimization of the products that the user pays more attention to in the user's comments but have poor performance,and form the product closed-loop management through continuous monitoring,so as to improve the market competitiveness of the products and further expand the market share of the enterprise.