BROWSE
Detail
Conference Paper
SEA-SF : Design of Self-Evolving Agent based Simulation Framework for Social Issue Prediction
Cited - time in 
Share 
Share
- Authors
- Joonyoung Jung, Euihyun Paik, Jang Won Bae, Dongoh Kang, Chunhee Lee, Kiho Kim
- Issue Date
- 2016-11
- Citation
- International Conference on Human and Social Analytics (HUSO) 2016, pp.67-70
- Language
- English
- Type
- Conference Paper
- Abstract
- Simulation is the imitation of the operation of a real-world process or system over time. Actual real world expectation is expensive and impossible because the modern society is complex and various. Therefore, simulation can be carried out to take proper measures for the problem which may be happened in the future. Agent based model (ABM) models each individuals and interactions among them. ABM mostly defines behaviors based on rule. However, ABM simulation has the weak point that simulation error is accumulated. If long term simulation is conducted, the simulation result will be highly inaccurate because of error accumulation. To overcome error accumulation, the model should be reconfigured using the real data recursively. In this paper, we propose the self-evolving agent based simulation framework (SEA-SF). The SEA-SF is consisted of data management, change recognition, model evolvement, ABM reconfiguration, user interface and ABM simulation environment. The SEA-SF should mitigate the long-term simulation error. Therefore, the SEA-SF performs change recognition between real data and simulation result. And then autonomously, it updates model parameters or the model configuration to increase accuracy of simulation. The proposed framework can be applied to solve the social issue problems because the social issue problems are happened through a long period. Therefore, the social issue simulation, such as the house policy and supply, can be performed using the proposed SEASF.
- KSP Keywords
- Data Management, Error accumulation, Long period, Long-term simulation, Model configuration, Model parameter, Over time, Real data, Real-world, Simulation Environment, Simulation framework