| Abstract |
| In the automotive aftermarket, user experience (UX) is typically measured using numerical ratings; however, such rating-based assessments rarely capture the emotional tone and contextual dissatisfaction expressed in user reviews. We introduce a mixed analytical framework that combines numerical ratings with text-based sentiment analysis and topic modeling to comprehensively interpret UX in the automotive aftermarket. To examine how numerical ratings, sentiment scores, and latent topics reflect user experience, we analyzed online review data from the four major aftermarket categories. At the aggregate level, sentiment scores were positively correlated with ratings. However, at the category level, this relationship became weak or unstable when rating distributions were highly concentrated. Topic modeling also showed that reviews with similar ratings differed in their experiential contexts (e.g., usability problems, functional inconvenience, login failures, and connectivity issues). Specifically, usability and system instability were rated lower and received more negative sentiment, while satisfaction-oriented topics received higher ratings and positive sentiment. Regression-based analyses showed that sentiment and topic-derived features can complement rating-only evaluation, indicating that automotive aftermarket UX should be interpreted through numerical ratings and the emotional and contextual information captured in review text. |
|
|
| Key Words |
| Automotive Aftermarket, Online Review Analysis, Sentiment-Based Text Analysis, Topic Modeling, Satisfaction Prediction Model, 자동차 애프터마켓, 온라인 리뷰 분석, 감성 기반 텍스트 분석, 토픽 모델링, 만족도 예측 모델 |
|
|
 |
|
