Consumption Demand Analysis Of Experiential Commodities Based On Text Mining

With the rapid development of the Internet era,more and more consumers choose to purchase goods online.Although online shopping has the unique convenience of offline shopping,online shopping also has obvious disadvantages,especially for the purchase of experiential products,whose purchase perception depends more on subjective attributes such as personal taste and product experience,while online shopping Shopping is likely to result in disproportion between consumers’ expected utility and actual utility due to differences in consumers’ cognition of products.The current research on online reviews of experiential commodities is mainly biased towards the usefulness of product reviews and the logistics service research of product negative reviews,and the research on the utility difference of experiential commodities is slightly less.This thesis adopts the method of combining text mining and KANO model,selects the representative of headphones as a carrier,constructs a consumer sentiment index system by studying the real needs and use satisfaction of users in the process of online shopping,finds out the factors affecting their satisfaction,analyzes the characteristics of user behavior combined with user feedback,proposes corresponding improvement strategies to improve consumer experience,and takes this factor as consumer attention,analyzes the type of product feature demand,and excavates product demand points.Finally,we provide corresponding countermeasures for suppliers.This thesis first collects data,crawls a total of 24,431 valid comments,and uses python to complete text deduplication,compression and word removal,short sentence deletion,part-of-speech manual tagging and Chinese word segmentation preparation.Then use the high-frequency words to statistically analyze the customer’s emphasis on each feature of the product,get the side that consumers care about most,and present the word cloud intuitively based on the results,and analyze the topic information of the review samples in different categories,because This information can only respond to consumers’ attention in terms of word frequency,so it may lead to inaccurate attribute extraction.Using the ROSTCM6 software package to conduct semantic network association analysis,we can obtain the internal correlation between various influencing factors,and use this as a basis,establish LDA topic models for positive reviews and negative reviews respectively,and the function of mining topic information is to express the focus of consumers’ attention through the correlation degree of feature words,and construct the subsequent consumer sentiment index system by mining topic information.Secondly,using the results of the previous LDA topic model to improve the emotional lexicon and feature lexicon and construct sub-dimensions,then use Boson NLP to verify the accuracy of the data set and score the sentiment of the feature words of the comment text data,and divide the dimensions to evaluate the product The scores of each indicator are counted.Finally,emotional analysis is carried out in combination with the emotional word score and product characteristics to obtain the emotional distribution of consumers for the product;for the analysis of product demand,in addition to considering the negative evaluation of the product,the proportional weight of product feature indicators and Emotional factors also need to be taken into consideration.Among them,the result of emotional analysis is an important symbol reflecting the degree of consumer concern,and it plays a dominant role in considering product demand analysis.Finally,the requirements classification method combining the KANO model and feature words is adopted.Since the attention between different themes does not belong to the same magnitude,the demand type is classified for different topics,the demand type is classified by the emotional merit score and the attention score,and the emotional words with degree adverbs are weighted,and the KANO demand model classification and the negative evaluation proportion of feature words are combined to calculate and rank the demand priority,and the factors affecting user demand are found out by analyzing the requirements of each level.According to the degree of user demand after sorting it,it is found that in the product theme in front of the sound effect,battery life and convenience,etc.,in the forefront of the logistics theme are complaints,returns,personnel communication attitude,etc.,at the same time to sort the characteristics and the product characteristics of the demand type one by one analysis,so headset online sales,to find the product market,targeted to improve the wearing experience of the product itself,upgrade the after-sales service experience level,form the product itself characteristic services,to meet the basic needs of users.On this basis,it is necessary to improve the after-sales service level of the product,and improve user satisfaction through the optimization of these expectations and attractiveness.