Examination of Best-time Estimation using Interpolation for Geotagged Tweets

Various studies have been conducted to analyze social media data in real time and to extract events in the real world. A benefit of analysis using data with position information is that it can accurately extract the event from a target area to be analyzed. However, because the proportion of data with position information in social media data is small, the amount to analyze is insufficient in almost areas. In other words, the problem indicates that we cannot fully extract most events. Therefore, efficient analytical methods are necessary for accurate extraction of events with position information, even in areas with few data. For this research, we conducted an experiment using information interpolation to estimate the times for biological season observation using tweets with Twitter location information. Then we evaluated its effectiveness. Herein, we explain results obtained using interpolated information and analysis of cherry blossoms in Japan in 2016.