Recently, a lot of people use Social Networking Service (SNS) to report their situation and do some findings around them. Especially, “twitter” is one of the easy ways to report some information for followers, and “twitter” users could send information and personal situations around them to “twitter” platform in any time. The amount of “twitter” data might be a lot more than the existing statistical data since some information are being uploaded at any time the users find it. Therefore, the numbers and the contents of “tweets” are different depending on the scale of the event, and a relevant key word is reported to “twitter” by the users. For example, when the railway operation stops, “twitter” users report that situation for the followers. The contents of the “tweets” are railway accident information, recovery time and complaint to railway company/staffs and so on. Then the duration of suspension time of the railway operation can be forecasted using the “twitter” data of the railway users. In this study, number and contents of “tweet” are analyzed under accidental conditions which are a small railway accident and a railway accident in which a lot of people get influenced. Furthermore, the authors have developed suspension time forecasting model using “twitter” data, and the forecasted time is verified by comparing with the existing statistical data which is utilized by the Ministry of Land, Infrastructure, Transport and Tourism in Japan.

railway, Social Networking Service, twitter, forecasting model