Abstract: Social media, including micro-blogs such as Twitter, have become main channels to communicate and share public social opinion among people. The opinion on nuclear power is an important social issue because nuclear plant construction needs national consensus. This paper proposes an opinion mining approach to monitor public sentiments on nuclear related issues using tweets on Twitter. The proposed process consists on (1) Crawling related tweets (2) Text preparation (3) Sentiment dictionary construction, and (4) Sentimental scoring. Based on experiment using nuclear related tweets in Korean between 2009 and 2013, we verify the usefulness on the proposed approach and confirm the changes on national opinion on nuclear generation depending on critical events such as Fukushima Daiichi nuclear disaster.

Keywords: public opinion mining, sentiment analysis, nuclear power.