Data Driven Spatio-Info Network Modeling and Evolution with Population and Economy

Abstract

Spatial interaction is the process that individuals interact with each other at different geographical locations. It attracts much research interests for the increasing data and applications related to spatial interaction. In this paper, a method is proposed to construct the spatio-info network with the dataset from WeChat. The correlation between human factors and statistics characteristics of the network is analyzed and confirmed, and then, the gross domestic product (GDP) and demographics are integrated into gravity model to model the spatio-info network. The likelihood method is used to solving the parameters and evaluates the four models; it is found that the GDP-GDP-distance (GGD) and population-population-distance (PPD) are similar and much better than the other two models. Finally, topological characteristics and community structure of the evolution network are analyzed to evaluate the models. It is found that evolution networks of the two models are almost consistent to origin network, and PPD models are better. It is concluded that the gravity model and human factors can be used to model the spatio-info network. This paper can be used to predict the communication amount of different regions in online social media dynamically. Naturally, this will help the mobile communication infrastructure construction, especially for a new generation of technology, such as 5G, or for regions with poor infrastructure. In addition, it will also help the software service providers configure server and advertising resources.