Research On Users’ Preference For Domestic Sports Shoes Based On Text Mining

Under the influence of the national internal circulation strategy and the "Xinjiang Cotton Incident",more and more consumers are buying domestic sports shoes through online shopping.The main market share of domestic sports shoes is in China,among which the three most representative brands are Anta,Li Ning and Xtep,and their market share is as high as 26.8%.This is mainly due to the huge transaction volume of e-commerce platforms.Each brand has a large amount of comment information.Comments are an intuitive reflection of consumers’ attitudes and are valued by platforms and manufacturers.Platforms and vendors can discover their own strengths and weaknesses through comments,update management strategies in time,and gain a leading position in the competition.Due to the huge amount of review data,chaotic structure,and serious colloquialism,it is difficult to obtain useful information from reviews,and the cost is high,and there has been no relevant research on reviews of domestic sports shoes in the past.Therefore,it is of great practical significance to efficiently analyze the domestic sports shoe review text to extract the target information.This article first introduces the research background and significance of domestic sports shoes e-commerce reviews,summarizes and summarizes the relevant theories of existing text analysis,and presents the research methods of this article.Then it expounds the related technical theories of network data collection,text sentiment analysis and topic extraction.Next is empirical analysis,using Python crawlers to obtain comment data,deduplication of the text,etc.,and finally obtain 41,230 data,and then use the word2 vec training word vector for subsequent classification of the cleaned text,respectively using the naive Bayes algorithm.FV-SA-SVM,CNN-BI-LSTM algorithm and CNN-BI-LSTM-Att algorithm are used for classification.The evaluation index of classification performance is mainly F1 value.Among them,CNN-BI-LSTM-Att algorithm has the best classification effect.The F1 value is 0.989.Finally,the LDA theme model is used to extract the theme of each brand from the classified review data,and the image analysis of each brand’s shoes is carried out.When the same theme word appears in both positive and negative theme words,it is judged by weight.The positive characteristics of the Anta brand are: shoes are breathable,comfortable to wear,and of good quality.The main negative characteristics are: rough details,poor customer service attitude,and inadequate after-sales protection.The positive characteristics of the Li-Ning brand mainly include: domestic production,good workmanship details,good celebrity publicity,and the main negative characteristics: low cost performance,slow logistics,and discrepancies between promotional pictures and actual products.The positive characteristics of the Xtep brand mainly include: high cost performance,a wide range of users,and simplicity.The negative characteristics include poor quality,rough workmanship,and poor materials.Based on the previous empirical analysis results,suggestions are made to the JD platform to improve the packaging of logistics and prevent secondary sales of shoes.The suggestion for each brand is that the Anta brand strengthens the handling of details,improves the attitude of customer service,and strengthens after-sales service.The Li-Ning brand achieves high quality and low price.The logistics of the Jingdong platform will be faster.The picture promotion and the physical product Match.Xtep brand uses better materials,improves workmanship,makes shoes more breathable,and improves product quality.