Decoding public sentiments on energy transition in Japan's three major metropolitan areas: Social media analysis using machine learning

This study employed a systematic analysis of social media data from Japan's three major metropolitan areas (Nagoya, Osaka, and the Tokyo metropolitan area) to investigate public sentiment regarding the energy transition. A social media analysis framework was developed for measuring public sentiment regarding energy transition. Initially, a Japan-specific energy transition-related dictionary was created. Subsequently, data were extracted from social media platforms in a scientifically rigorous manner. Advanced text mining techniques were employed to find the most suitable machine learning models and construct indicators of positive and negative sentiment related to the energy transition. The study ultimately sought to quantify the temporal fluctuations in public sentiment. In light of these findings, particular attention was paid to the relationship between sentiment and external shocks. An innovative approach was taken in applying a structural breakpoints test to analyze the relationship between regional sentiment fluctuations and changes in energy prices, while also considering the potential impact of extreme disasters on public sentiment. These findings suggest that policymakers should increase transparency and improve communication with the public during periods of energy price volatility, clearly explaining the reasons for price changes and response strategies. In addition, in areas where interest in energy transition is low, policy advocacy and education efforts should be strengthened to promote broader and stronger public support. This study not only provides a deeper understanding of public sentiment regarding energy transition but also offers new perspectives for energy policy formulation, regional adaptation, and adjustment with scientific evidence. © 2025 Elsevier Ltd