Scholarlink

<p>With the rapid growth of cross-border e-commerce, logistics service quality has emerged as a critical determinant of consumer satisfaction and platform competitiveness. Traditional evaluation methods relying on subjective judgment and limited indicators fail to fully capture consumers’ actual experiences. This study proposes a multidimensional evaluation framework integrating deep learning-based sentiment analysis with survey data. Employing a RoBERTa-BiLSTM-Attention model alongside Latent Dirichlet Allocation (LDA) topic modeling and Importance-Performance Analysis (IPA), we assess logistics service quality across three platform types. Results reveal that Platform A (comprehensive e-commerce) excels in delivery timeliness but lacks information transparency, where targeted improvements could reduce consumer anxiety and boost repurchase rates; Platform B (vertical e-commerce) performs well in cost and speed but requires enhancements in delivery security and return services to lower perceived transaction risk and expand its high-value product market; Platform C (cross-border community group-buying) shows strengths in delivery efficiency but faces dissatisfaction in customs clearance and logistics tracking, with improvements expected to shorten delivery cycles and strengthen consumer trust in the group-buying model. These findings offer actionable insights for platform operators, policymakers, and researchers engaged in advancing consumer experience and platform governance in cross-border digital commerce.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A multidimensional evaluation of cross-border e-commerce logistics service quality based on deep learning sentiment analysis

  • 期刊论文

  • 发表日期：

    2026年5月19日

查看全文
  • Te Li,
  • Hadrian Nichols

摘要

With the rapid growth of cross-border e-commerce, logistics service quality has emerged as a critical determinant of consumer satisfaction and platform competitiveness. Traditional evaluation methods relying on subjective judgment and limited indicators fail to fully capture consumers’ actual experiences. This study proposes a multidimensional evaluation framework integrating deep learning-based sentiment analysis with survey data. Employing a RoBERTa-BiLSTM-Attention model alongside Latent Dirichlet Allocation (LDA) topic modeling and Importance-Performance Analysis (IPA), we assess logistics service quality across three platform types. Results reveal that Platform A (comprehensive e-commerce) excels in delivery timeliness but lacks information transparency, where targeted improvements could reduce consumer anxiety and boost repurchase rates; Platform B (vertical e-commerce) performs well in cost and speed but requires enhancements in delivery security and return services to lower perceived transaction risk and expand its high-value product market; Platform C (cross-border community group-buying) shows strengths in delivery efficiency but faces dissatisfaction in customs clearance and logistics tracking, with improvements expected to shorten delivery cycles and strengthen consumer trust in the group-buying model. These findings offer actionable insights for platform operators, policymakers, and researchers engaged in advancing consumer experience and platform governance in cross-border digital commerce.