Research On Multi-dimensional Travel Activity Pattern Recognition In Location-based Social Networks

The development of mobile internet,geospatial information technology,Location-Based Service(LBS),and intelligent transportation technology provides new data and theoretical support for obtaining the information of urban residents’ travel activities and analyzing their travel activities.This research hopes to apply the location-based social network(LBSN)to analyze residents’ travel behavior.Based on the LBS trajectory data,residents’ daily activity type and travel mode structure are mined from multi-dimension,providing a reference for planning and managing urban traffic.In this paper,the theory of trip pattern recognition,LBSN,individual trip activity chain,and individual trip space-time distribution are applied.Based on studying urban residents’ daily travel and activity,three pattern levels based on LBSN hierarchical network are constructed,and the pattern characteristics of three angles of spatial location,temporal-spatial activity,and trip activity chain are analyzed.Using SDK track data,combined with the LBS location data acquisition principle,build track data cleaning algorithm.Divide urban residents’ daily behavior into the resident phase and trip phase,and the residential area in the space is identified by the CLIQUE space clustering algorithm based on density grid.The resident points in the trajectory are identified according to the static condition,the activity of the resident point is matched with the POI function attribute.For the sequence structure of resident activities,12 patterns among the function-oriented activity pairs were found using the theme-generated LDA model.JS divergence was used to measure the similarity of any two resident activity chains.The travel activities of individuals are represented by space-time path and space-time prism,and compressed linear reference technology CLR represents the dimensionality reduction.This paper analyzes the accessibility of individual space-time activity under the restriction of spacetime,activity type,and traffic network and obtains four modes of space-time interaction of individual activity assisted by ICT technology.In the single-day complete trip activity chain structure,four trip patterns are constructed based on a single trip.By comparing the SDK path data with the path planning API of Amap,the corresponding path and traffic mode can be obtained with more than 70% accuracy.The complete travel activity chain and the corresponding phase of the related activities and traffic patterns are identified from the trajectory data.We used SDK track data in Shenzhen and based on LBS location information for empirical analysis.The pretreatment of the original data,such as cleaning,and visualization,is applied to resident point recognition,activity chain pattern recognition,individual daily travel activity chain recognition,and corresponding appraisal.The results represent that the LBSN hierarchical research method has a good performance for travel activity pattern recognition using LBS trajectory data.