2018 Volume 16 Issue 2 Pages 283-294
From the viewpoint of prevention of crowd accident, it is important to early grasp how much and where local/sudden congestion by the crowd is occurring. In this paper, we discuss the possibility of quantifying the crowd congestion degree based on the expressions in text information posted to SNS, which is excellent in immediacy. Firstly, 52 expressions for the degree of crowd congestion and 100 photographs of congestion situation are collected from the actual Twitter data posted within 24 hours after the Great East Japan Earthquake occurred. Secondly, we ask the respondents of the Web-based questionnaire survey to evaluate the degree of congestion recalled from each linguistic expression on a 10 level scale (1 [low] – 10 [high]). In addition, the respondents are asked to choose a linguistic expression suitable for expressing the crowd congestion degree in each situation presented by one of the photographs. Based on these results, we score and rank 52 expressions and 100 photographs, and demonstrate the usefulness of text information on SNS in estimating crowd congestion degree.
