Community Discovery Based On Public Transportation Data

With the development of communication technology,wireless networks and data mining,urban travel trajectory data has become an indispensable support for depicting human society and activities at the microscopic scale.Urban travel trajectory data can not only display the mobile information of different objects in real time,but more importantly,through the analysis and mining of these data,the hidden knowledge and patterns can be found,so as to better serve people's daily life.In particular,the discovery of valuable information from massive urban public transportation data has attracted wide attention from scholars.This thesis takes public transportation data of Chongqing in March 2017 as the research data,and conducts research on the communities existing in the public transportation network.Firstly,this thesis establishes a pre-processing method for urban public transportation data,which uses this method to clean data and prepare data for public transportation card data,public transportation line data and public transportation stop data.Secondly,this thesis proposes a method for calculating the boarding based on transfers,and utilizes the travel chain to infer the alighting.Then,based on commuters,this thesis proposes a community discovery method for social networks.Finally,based on Chongqing's public transportation data,this thesis implements a social recommendation platform “Travelpooling”.More specifically,the main contributions of this thesis are as follows:(1)In view of the problem of location similarity and low computational efficiency for masses public transportation stops,this thesis proposes a grid-based public transportation road network division method,which uses grid to replace the public transport stops,so as to simplify the calculation for each stop and improve the computational efficiency.The method calculates the density of public transportation stops in each grid,and regards the corresponding grid granularity as the optimal grid granularity in the case of the largest number of dense grids.The experimental results show that the actual distance of the stops within the divided grid is similar,and the grid replacement effect is significant.(2)Aiming at the problem that flat fare bus does not record the boarding and the alighting,this thesis proposes a method to calculate the boarding and the alighting based on transfer and travel chain.The method comprises the clustering analysis of public transportation card data,determining the passengers of the boarding according to the time interval,determining the boarding by using the transfer,complementing the passengers' trips based on the travel chain,and visualizing the passengers' trips.The experimental results show that the method does not rely on external data,and can achieve higher accuracy.(3)Based on the passenger's travel,this thesis proposes a community discovery algorithm and applies it to social recommendation.Firstly,the algorithm uses the regularity of commuter to measure the temporal,space and semantic similarity.Secondly,use the similarity for community discovery.Finally,use the results found in the commuter passenger community for social recommendations.The experimental results show that the community discovery algorithm proposed in this thesis has a good effect,and the commuter in the community have high similarity.(4)Based on the community discovery algorithm,this thesis develops a social recommendation platform based on the current work and life place.The platform first introduced the functions of friend recommendation,hot news,community sharing,annual travel report and basic information maintenance.Then,we implemented the platform based on the structure of “WeChat + WebService + Hadoop”.Finally,we deploy it on WeChat.