21st International Conference on Computational Science and Its Applications, ICCSA 2021, Virtual, Online, 13 - 16 September 2021, vol.12957 LNCS, pp.220-230
© 2021, Springer Nature Switzerland AG.Human trajectories provide spatial information on how citizens interact with the city. This information can explain the daily routine, economic and cultural cycle of citizens. In the last year, the pattern of human trajectories is expected to change over cities due to Covid-19. This paper investigates the change in human trajectories based on the social media data (SMD) before and after the Covid-19. This study aims to find out the differences between the years 2018, 2020, and 2021. Firstly, all accounts for each year are classified based on movement behaviors. Secondly, spatial distributions of the tweets in terms of classified accounts are visualized after the hierarchical clustering applied to each dataset. Lastly, the average step lengths (ASL) are calculated for each account and classified in terms of step length levels as no movement, neighborhood, district, inter-districts, inter periphery, and center, outbound. The number of tweets and distinct accounts decreased by 90% and 84% from 2018 to 2021. The decrease in the number of single tweeting accounts is 84%, it is 60% in stationary accounts, and 94% in moving ac-counts. The size of the spatial clusters also decreased for all types of accounts maps, however, some of the previously visited spatial points are disappeared while new ones appeared on maps of single tweeting and moving accounts. The ASL of moving accounts also confirms the human movement decrease. According to that, the max, mean, and median ASL decreased 22%, 13%, and 35%. Results point out outcomes vary in terms of accounts’ movement behaviors. This study is expected to contribute the measuring the pandemic impacts on human movement with SMD.