Text Mining Analysis of Radiological Information from Newspapers as Compared with Social Media on the Fukushima Nuclear Power Plant Accident

Abstract

After the nuclear meltdown incident at Tokyo Electric Power Company’s Fukushima Daiichi nuclear power plant in March 2011, citizens are increasingly concerned about the adverse effects of radiation on health. Perception and anxiety about risks have been worsened by inaccurate news reports and a lack of information about the incident. In this research, newspaper headlines from March 2011 to January 2012 on the Fukushima incident were comprehensively collected and analyzed using text mining. Extracted data were also compared to information transmitted via social media such as Twitter during the same period to examine the following four issues related to features of the two types of transmission, i.e., newspaper headlines and social media, and the relationship between information from the media and anxieties about the adverse effects of radiation on health. 1) Over 10 months after the nuclear meltdown incident, the information most frequently provided by newspapers was that related to radiation measurement and surveys but providing few articles directly related to adverse effects of radiation on health. 2) Newspaper headlines combined multiple topics in short transmitted text while, at the same time, the limited information transmitted by social media was exaggerated. 3) Newspapers indirectly explained the adverse effects of radiation on health using concepts such as “danger/risk,” “safety,” “anxiety/dissatisfaction,” and “security.” 4) Many articles used the term “danger/risk” in March 2011 before the degree of radiation exposure had been evaluated reliably.