What Is an Artificial Society?

An artificial society is a virtual society constructed inside a computer. If information about the people in the real world could be used, an artificial society could be built with ease. However, for reasons of personal data protection and privacy, using information about actual individuals is not permitted. For this reason, we conduct research on constructing artificial societies using only publicly available information, such as official statistics and open data. We have already succeeded in generating the attributes of individuals for all of Japan, and we are currently pursuing research toward realizing an artificial society in which individuals act, using generative AI.

Example Research Topics

• Developing a Method for Assigning Personalities to Synthetic Population Data for LLM-Driven Social Simulation
• Assigning National Pension Beneficiaries and Benefit Amounts to Synthetic Population Data
• Building Use Classification Using Deep Learning: Toward Improving the Accuracy of Address Information in Synthetic Population Data
• Synthesizing Prefecture-Level Microdata That Accounts for Municipal Statistical Tables

Integration with Data Science

To run a simulation, we first need to know what is causing a social problem and how it can best be improved. To this end, we analyze the current state of society using real data and artificial societies, and identify the issues to be addressed.

Example Research Topics

Mobility

• Estimating the Usage Status of Individual Bicycles Based on Bike-Sharing Data

Education

• Analyzing the Impact of Elementary School Consolidation on Children's Commutes to School
• Estimating the Number of Hidden Waitlisted Children for Childcare in Japan

Disaster Prevention

• Suitability Analysis of Evacuation Shelter Locations Using Synthetic Population Data

Healthcare

• Analysis of the Public Availability of AED Location Information Across All Municipalities in Japan
• Examining AED Placement Assuming Cardiac Arrest Anywhere Within a Facility

Evaluation through Social Simulation

We design policies and institutions to address the issues identified through data science, and evaluate their effects through simulation. We also explore ways of presenting the policies and institutions examined in our simulations, and even develop games, in order to communicate them.

Example Research Topics

Mobility

• Verifying the Introduction of Ride-Sharing to Secure Means of Transportation in Depopulated Areas
• Examining the Profitability of Demand-Responsive Buses through Simulation

Disaster Prevention

• Providing Information to Foreign Tourists during Evacuation Using Social Simulation

Healthcare

• Visualizing the Effects of an AED Delivery Support System for Cardiac Arrest at Home
• Developing a Serious Game for Learning the Characteristics of Nursing Care Facilities That Match Prospective Users' Needs