The Evolution of Public Perceptions of Automated Vehicles in China: A Text Mining Approach Based Dynamic Topic Modeling

Public attitudes and intentions are crucial for successful technological innovation, such as automated vehicles (AVs). To effectively advance the development and future evolution of AVs, it is crucial to comprehensively understand individuals’ perceptions of autonomous driving. However, traditional survey methods using structured questionnaires may limit respondents’ ability to express themselves freely. To address this limitation, we employ Natural Language Processing (NLP) techniques to analyze consumers’ opinions and attitudes towards AVs as shared on social media platforms. Through Python programming, we collected and analyzed consumer comments from leading Chinese social media platforms (Sina Weibo, TikTok) and automotive social media platforms (Autohome Inc.) between June 2020 and April 2023, totaling 120,486 comments. Leveraging advanced text mining techniques such as Dynamic topic models (DTM), sentiment analysis, and semantic network analysis based on Pointwise mutual information (PMI) algorithms, we investigate the evolution of public perception regarding AVs over the past three years. Our findings unveil a predominant negative sentiment towards AVs, with discernible shifts in sentiment coinciding with major AV-related social events. Furthermore, we explore the reasons behind users’ negative attitudes and identify potential factors contributing to the distrust of autonomous driving. These findings provide valuable guidance to public agencies, automobile manufacturers, and technology companies, enhancing their understanding of the adoption of AVs.