Answers are grounded in this paper and cite their sources.
In the context of the rapid increase in the market penetration rate of new energy vehicles, pain points such as range anxiety restrict the development of the industry. Existing studies mostly treat the entire review as a single sentiment calculation unit, ignoring the multidimensionality and complexity of consumer reviews, which leads to insufficient accuracy. Therefore, this study conducts text mining based on 10,981 owner reviews of the top 20 best-selling models on the Dongchedi platform. By extracting core features through TF-IDF and co-word networks, the LDA model is utilized to identify five major themes: overall vehicle evaluation, driving experience, car purchase decision, household comfort, and range anxiety. Combining SnowNLP technology for fine-grained sentiment quantification effectively avoids the bias where local emotions are masked by overall evaluations. The study finds that user sentiments exhibit significant polarization: there is high satisfaction with cabin comfort and overall vehicle performance, but the proportion of negative sentiments in the range anxiety dimension is relatively high. The research results accurately reveal the current technical and service shortcomings of the industry, providing decision-making support for car companies' optimization.
No takes yet. Share an insight, caveat, or question.
Runjia et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: