소셜 미디어 리뷰 기반 전기차 소비자 감정의 단계별 진화 분석 = Understanding Consumer Dissatisfaction and Acceptance of Electric Vehicles through Dynamic Topic Modeling and the Lifecycle Perspective

다국어 초록 (Multilingual Abstract)

As the global electric vehicle market expands rapidly, discussions among consumers regarding the technological performance, charging convenience, and policy environment of electric vehicles are becoming increasingly active. However, despite continuous improvements in market scale and technological conditions, user reviews on social media remain focused on negative topics such as battery safety risks, charging wait times, uneven charging station distribution, and price pressures. This indicates a distinct gap between market development and consumer perception. This study aims to uncover the structural characteristics and temporal dynamic evolution patterns of negative perceptions among electric vehicle consumers based on public discussion data, thereby providing a foundation for technological innovation, enhanced market acceptance, and policy formulation. The study utilised a total of 134,374 electric vehicle-related reviews from the Reddit platform between January 2023 and December 2024 as its data source. After cleaning for duplicate entries, non-English text, URLs, and invalid content, 69,969 valid reviews were ultimately analysed. This study employs text mining techniques such as sentiment analysis, Latent Dirichlet Allocation (LDA), and Dynamic Topic Modelling (DTM) to identify negative sentiment, extract thematic structures, and analyse trends over time. It also combines these with external events to perform validation over time. The results revealed that negative sentiment accounted for 16.8% of all reviews. Negative reviews contained key concepts such as “car”, “EV”, and “battery”, accompanied by emotional vocabulary like “fuck” and “shit”, indicating a coexistence of technical issues and emotional catharsis. Thematic consistency analysis identified five optimal themes: electric vehicle price competition and market strategy; subsidy and policy changes; range anxiety and energy consumption concerns; the impact of extreme weather on vehicle performance; and battery safety risks. DTM results revealed distinct temporal fluctuation characteristics for negative themes, with each peaking in different months and exhibiting periodic, concentrated spikes. High synchronisation between significant external events and peaks: for instance, Tesla's price reduction in January 2023, the price and subsidy policy adjustments in April 2024, and the fire at a Korean battery factory in June 2024 all caused a sharp surge in the intensity of specific themes. These findings demonstrate that policy changes, accident events, and market fluctuations act as triggers for negative sentiment diffusion. The intensification of negative perceptions is not random but reflects a structural response mechanism driven by external events. The research contribution manifests in three aspects. Firstly, it reveals that consumer complaints in digital spaces exhibit a structural evolution model based on the dynamic association between sentiment, topic, and event. Second, the findings provide emotional warning signals for policy formulation. An abnormal surge in the intensity of specific topics signifies a potential risk of negative sentiment diffusion, necessitating pre-emptive communication intervention. Third, the research confirms that battery safety and charging inconvenience are the most critical and recurrent negative factors. Consequently, technological enhancement, infrastructure development, and safety management represent the areas where enterprises and governments should prioritise investment. While this study focuses on negative reviews, it simultaneously analyses positive experiences, limiting a balanced understanding of consumers' overall perceptions. Furthermore, the causal impact of external events on emotional peaks has not been rigorously quantified; future research could introduce time-series causal models for further discussion. Overall, this paper demonstrates that negative comments on social media not only reflect user experiences but also hold significant value in encouraging innovation and guiding policy. The phased evolution of consumer sentiment provides a crucial perspective for understanding the development of the electric vehicle market.

목차 (Table of Contents)

• 1. Introduction 1
• 2. Literature Review 3
• 2.1 Research on Electric Vehicle Adoption 3
• 2.2 The Effectiveness of Incentive Policies and the Adoption of Electric Vehicles: A Literature Review 6
• 2.3 Propositional Hypotheses 9
• 2.4 Theoretical Background 11
• 3. Research Methodology 13
• 3.1 Research Framework 13
• 3.2 Data Collection 15
• 4. Analysis Result 18
• 4.1 Sentiment Analysis Results 18
• 4.2 Topic Modeling (LDA) 21
• 4.3 Dynamic Topic Modeling (DTM) 23
• 4.4 Topic–Event Linkage and Interpretation of Negative EV Discourse 25
• 4.5 Cross-Topic Discussion 31
• 5. Conclusions 32
• 5.1 Discussion 32
• 5.2 Implications and Limitation 34
• REFERENCE 36