[4Xin1] Poster session 2

[4Xin1-64] Evaluation of restraint policies and estimation of infections by MAS with social sensing

Takumi Sugiura¹, 〇Naoki Wakabayashi¹, Ryoya Watanabe¹, Fumito Ihara¹, Takahiro Higasa¹, Satoshi Kurihara² (1.Graduate School of Keio, 2.Univ. of Keio)

Keywords:Multi agent simulation, Machine learning, social sensing

We have developed a multi-agent simulation system for COVID-19, which is still infecting people, focusing on human behavioral patterns. We have estimated the number of newly infected persons and effects of measures against COVID-19 by determining behavioral parameters for the simulation.
However, these parameters are determined empirically based on real-life human flow data, and it has been difficult to determine the parameters quantitatively. In our analysis of social media for COVID-19, we showed that it is possible to estimate the willingness to go out and turnout using Twitter, a representative social media platform.
Therefore, in this study, we consider that there is a relationship between the behavioral parameters of the simulation and the emotions. We construct a model that explains the behavior parameters using the emotions estimated from Twitter. Then, we evaluated the restraint policy by running a simulation reflecting the public reaction to COVID-19 and estimating the number of infected people.