Sentiment analysis of nuclear waste on Twitter and Instagram data

—This study investigates public opinion and sentiment toward Japan's Fukushima nuclear wastewater discharge as expressed on social media, specifically Twitter and Instagram. The goal is to perform sentiment analysis to provide stakeholders with a reliable foundation for informed decision-making. The main issue of this study is the absence of a comprehensive analysis of public opinion and sentiment on social media regarding nuclear wastewater discharge from the Fukushima Daiichi Nuclear Power Plant. Thus, capturing and understanding these sentiments is essential, as they provide crucial insights into public opinion. Hence, a dataset of 12,195 posts from January to March 2024 was collected using Python’s Tweepy and Instagrapi libraries. Data preprocessing involved cleaning non-English text, hashtags, usernames, URLs, and emojis, followed by stop word removal. Key topics such as "nuclear," "waste," and "Fukushima" were identified through LDA topic modeling, focusing on posts highly relevant to the issue. Sentiment analysis was conducted using the BERT (uncased) model, categorizing sentiments into "very positive," "positive," "neutral," "negative," and "very negative." VADER was also applied to validate and compare the results, enhancing the robustness of the sentiment classification. Findings show that negative sentiment consistently outweighs positive sentiment, indicating an overall unfavorable public opinion toward the discharge. This analysis provides policymakers and stakeholders with valuable insights into and guidance for addressing global concerns about Japan's nuclear wastewater strategy. Future studies could include additional social media platforms and incorporate geolocation data to analyze regional sentiment trends more precisely.

Text 127756.pdf - Published Version Available under License Creative Commons Attribution Share Alike. Download (1MB) Official URL or Download Paper: https://ijaseit.insightsociety.org/index.php/ijase...

Actions (login required)

View Item