Description Of Urban Area Function And Analysis Of Human Mobility Based On Sina Weibo

In recent years,with the maturity of perception technology and mobile internet,data such as traffic flow,meteorological,road network,point of interest,mobile trajectory and social media have emerged in the city.These heterogeneous big data provide the data foundation for the construction of smart cities.Meanwhile,the tremendous improvement of modern computer computing power and the maturity of parallel computing make it possible to solve various problems caused by the rising population in urban development by mining and analyzing these massive data.The concept of urban computing is gradually emerging.The description of urban area function and analysis of human mobility are two popular applications in urban computing.This paper studies these two applications based on Sina Micro-blog checkin data.Delineating urban functional areas is one of the long-standing questions in urban studies and planning.From the perspective of urban buildings,this paper extracts time series data of buildings based on check-in data.According to the assumption that social media activities in these buildings with similar functionality not only have similar spatiotemporal patterns but also are strictly correlated to the temporal information.We propose a novel Matching Time Series(MTS)method to calculate the distance of the time-series data and use this method to modify the least squares twin support vector machine(LST-SVM)to establish multi-classification models for workday and weekend respectively.Our models can accurately identify the function of buildings in urban areas.Then we combine the adaptive DBCSAN clustering algorithm and the Thiessen polygon method to propose a novel process for describing the functional distribution of urban areas.Compared with the actual situation of the study area,the proposed method has higher accuracy in identifying urban building functions,and can dynamically and clearly describe the functional distribution of urban areas through the data of different time periods.For the study of urban human mobility,we propose a model for predicting the user’s next moving position based on user activity data.This model combines the adaptive DBSCAN and K-medoids hybrid clustering algorithm,the Markov chain model and the activity category detection method,and can reduce the range of prediction space and further improve the accuracy of predicting the next position.Through relevant experimental verification,we not only obtain relatively high activity category prediction accuracy,but also could predict the users’ next arrival location conveniently and mobile trends.Then this paper uses Origin-Destination(OD)analysis to study the human mobility between urban areas by mapping user movement trajectories to city areas.And we have established an urban inter-regional human mobility model based on the theory of directed graphs and degree centrality in graph theory,and quantitatively assessed the importance of each area.The study results can effectively help urban decision makers understand the city and allocate resources.