Research On Prediction For Co-travel Relationship Social Network Of Passengers

Co-travel passengers relationship,as a special social relationship,has important research significance and application prospects in the field of civil aviation.Civil aviation passenger peer relationship network prediction aims at link prediction of a homogeneous network with civil aviation passengers as nodes and passenger common travel relationships as edges.For airports,mining the passenger relationship can improve accuracy of the airport’s black-and-white grading security check,which will greatly enhance the airport’s overall operating efficiency.For airlines,by building a complete co-travel passengers network,better routes can accurately be recommend for passengers.For the passengers themselves,the prediction of the traveler’s peer-to-peer relationship network can also greatly help protect the personal safety of the passengers.Due to the large number of passenger nodes and extremely rare edges in the co-travel network of civil aviation passengers,the network is highly sparse,and most of the existing social network link prediction methods are based on the similarity algorithm to analyze and predict the relationship strength between dense graph nodes.These prediction methods are not suitable for highly sparse civil aviation co-travel passengers network link prediction.Therefore,this paper first proposes to convert the limited hash fields into network data on the basis of a large number of reservation records of civil aviation passengers to build a co-travel relationship network of passengers.Then,in this network data,the relationship features between the passenger pairs are deeply mined and extracted and refined.The extracted passenger peer relationship features are feature-vectorized as a prediction model,and the classifier is used for training and prediction.The accuracy rate proves the effectiveness and applicability of feature extraction.Based on the characteristics of passenger peer relationship,this paper further studies the dynamic link prediction for co-travel network of passengers.The relationship prediction within the co-travel network that changes with time is called the future link prediction for co-travel network of passengers.In this paper,we use five kinds of centrality indicators to perform a centrality analysis on the sample data.On the basis of the characteristics of the characteristics of the co-travel relationship of passengers,the common neighbor topology characteristics of the co-travel network and the hierarchical community characteristics of the co-travel network are extracted as three types of characteristic attributes.Based on the concept of time window,these three types of features are extracted as observation windows in a certain period of time,and the next time period is used as a prediction window to establish a prediction model to predict the future link of the co-travel relationship network of passengers.Experiments show that the performance of each of the three characteristic attributes has a different degree of positive impact on the prediction results of the peer network,and the prediction results of the prediction model formed by the combination of the three characteristic attributes perform well.The rationality and applicability of the extraction.