Research On Quality Evaluation Of Fresh E-Commerce Logistics Service Based On Deep Learning

With the growing online economy and the increasing consumer need for online fresh food shopping,fresh food e-commerce is expanding rapidly.Fresh products,due to their perishable and deteriorating nature,place higher quality requirements on logistics.Nowadays,consumers leave a constant stream of evaluation data on various online shopping platforms,most of which are real and objective.It is a more objectively approach to grasp consumers’ emotional attitude and core concerns in this area by analyzing consumers’ evaluation of fresh food logistics services and to address the pain points of fresh food e-commerce logistics services symptomatically.Previous studies have also proved the feasibility of this method,but the method to analyze these massive evaluation data efficiently and accurately has been the focus and difficulty of research by experts and scholars.This article further explores the evaluation method of fresh produce e-commerce logistics service quality based on deep learning sentiment analysis technology,and provides a reference basis for fresh produce e-commerce to improve logistics service quality.In this thesis,I study the quality and problems of fresh e-commerce logistics services through deep learning sentiment analysis technology.The whole research framework is executed in the order of first crawling the evaluation data of fresh food on e-commerce websites using python crawler technology and doing sentiment analysis pre-processing on the data,including data cleaning,word separation and deactivation;at the meantime,according to the classical service model and former research results and integrating the logistics features of fresh food e-commerce,the fresh food e-commerce logistics service quality evaluations dimensions and indicators are structured;then,the TF-IDF algorithm is used to extract the text keywords and artificially filter the logistics keywords based on the conversion of words into quantitative representation,and the word vector cosine similarity algorithm is used to extend the logistics keywords and extract the logisticsrelated evaluation data based on this matching;then some of the evaluation data are manually labeled with sentiment labels(0-unsatisfactory,1-satisfactory)to form the training and validation sets,which are used to fine-tune the sentiment classification model RoBERTa-wwm-ext suitable for Chinese text classification,in order to be more suitable for this research task,the sentiment classification is performed on a large amount of logistics evaluation data without sentiment labeling,based on the results,the current user satisfaction and core concern indicators of the fresh food e-commerce company’s logistics services are analyzed.So as to investigate the quality and problems of the logistics service,and give relevant suggestions for the existing problems.This lab aims to study the quality and problems of logistics services and give related suggestions for the problem,in order to supply some research referents for the developing of e-commerce.The main object of this experiment is the more than 300,000 fresh food review data of JD Mall,and after the appeal of the emotion classification process,four fresh food ecommerce logistics evaluation dimensions(quality assurance,reliability,responsiveness and empathy)and emotion classification results were obtained.Based on the sentiment classification results,the current development of JD fresh produce logistics service was obtained.The concern of quality assurance,reliability and responsiveness were much higher than that of empathy,and the dissatisfaction level of these three dimensions was relatively high.Only by improving consumers’ satisfaction in all dimensions can we increase their trust in this platform,and the fresh food business can develop more healthily and sustainably and have the opportunity to create more economic and social value.