Optimization of Population Mobility Model of Artificial Society Based on Big Data Analytics

Abstract

It has become a promising research topic that using the Artificial Society (AS) solves social governance problems such as emergency management, environmental protection, and urban planning. The unceasing evolution of the AS requires high demands for the accuracy of the basic models in AS. With the development of the IoT and data science, researchers can refine the models in AS by contrasting the big data harvested from real society. This paper aims to improve the accuracy of the population mobility model in AS based on the analysis of the real-world dataset about urban population mobility and the comparison between the real-world data and the data generated from AS. To this end, firstly, the datasets are processed and analyzed by Spark. To improve the efficiency in this phase, we design the RDD structure for population mobility data and propose a novel query method, named partition traversal method. Then, the population mobility patterns are analyzed and compared, which are extracted from the datasets of real-world and AS, respectively. Based on the diversity between the two patterns, the optimization scheme of the population mobility model is finally proposed by adjusting the workplace selection. This study is an important reference for the modeling and optimization of population mobility in AS.