Mining Social Media to Understand User Opinions on IoT Security and Privacy

Internet of Things (IoT), as an emerging technology, has attracted lots of attentions in recent years, but also faced criticism regarding its security and privacy. Some user interviews have been done in the past to understand the security and privacy concerns of end users on IoT devices, but they were limited in the number of participants. To better understand user opinions on IoT security and privacy, we conducted a larger user study using Twitter and Reddit data. We collected more than 6 million tweets and reddit submissions using query search keywords such as drone security, smart camera privacy and etc and labeled them using flair pre-trained en-sentiment model. We performed sentiment analysis using BERT, a transformer-based model, and a neural network classifier. Our sentiment analysis results have shown that Twitter users are more positive towards IoT security and privacy, whereas Reddit users are more negative. On certain categories such as smart refrigerator, smart TV, drone, speaker, voice assistant, fitness tracker, and smartwatch, both users express negative sentiment. We also conducted a longitudinal study to understand how user opinions change over time. A continuous upward trend has been identified for both positive and negative sentiments on Reddit users. We further analyzed users’ specific concerns through topic modeling particularly related to smart lock, smart camera, and drone.