The Research And Implementation On Ranking The Aviation Passengers’values Based On Social Network

With rapid development of the economy, more and more people choose to travel by air, which serves as a quick and convenient way of transportation.In recent years, the number of civil aviation passengers increased year by year, which is a large part among domestic passengers. In the context of fierce market competition, discovering high-value passengers and offering them precision marketing and personalized services constitute some of the important links in marketing strategies of airline companies, which also would bring us more benefits and achieve a win-win. At the present stage, the calculation of high-value passengers is based only on passengers’ historical flight data, such as flight frequency, mileage, cost and so on.Actually, to measure the marketing value of a customer, we need not only to emphasize on the personal historical expenses, but on the potential future value. The historical value of customers may be attained based on the personal historical records of travel, and the potential value of customers could be deduced from the networks of passengers.In this article, we construct the co-travel networks of civil aviation passengers by analyzing and processing the data including the records of reserving seats and taking planes, gathered from the airline companies. The vertices in the networks denote the passengers, the edges denote the co-travel relation and the stress of ties between passengers are represented by the weights on the edges. The co-travel networks of passengers sort of reflect the real social connections between passengers, which is beneficial for measuring the potential value of passengers accurately.In this paper, we thoroughly studied the model of measurement on value of nodes, which are important organic parts in the co-travel networks. And then, we designed the ranking algorithm PassengerRank to measure the value of co-travel networks, considering the stress of ties between passengers. Finally, we proposed a model, which synthesized the value of individuals and that of networks.At last, we did experiments on the data set of passengers, which is mentions before, analyzed the performance of this algorithm and made comparison with other sorting algorithms. The experimental results showed that PassengerRank performed better when compared to the other sorting algorithms, even including the original PageRank and RFM model.