Sentiment Matters: An Analysis of 200 Human-SAV Interactions
This work addresses the need for effective conversational interfaces in SAVs, providing incremental insights for researchers and designers in transportation systems.
The paper tackled the problem of understanding human-Shared Autonomous Vehicle (SAV) interactions by introducing a dataset of 200 interactions, and found that response sentiment polarity is a key predictor of SAV acceptance and service quality, with zero-shot LLM prompts outperforming traditional methods in sentiment analysis.
Shared Autonomous Vehicles (SAVs) are likely to become an important part of the transportation system, making effective human-SAV interactions an important area of research. This paper introduces a dataset of 200 human-SAV interactions to further this area of study. We present an open-source human-SAV conversational dataset, comprising both textual data (e.g., 2,136 human-SAV exchanges) and empirical data (e.g., post-interaction survey results on a range of psychological factors). The dataset's utility is demonstrated through two benchmark case studies: First, using random forest modeling and chord diagrams, we identify key predictors of SAV acceptance and perceived service quality, highlighting the critical influence of response sentiment polarity (i.e., perceived positivity). Second, we benchmark the performance of an LLM-based sentiment analysis tool against the traditional lexicon-based TextBlob method. Results indicate that even simple zero-shot LLM prompts more closely align with user-reported sentiment, though limitations remain. This study provides novel insights for designing conversational SAV interfaces and establishes a foundation for further exploration into advanced sentiment modeling, adaptive user interactions, and multimodal conversational systems.