Abstract: This paper used the map-reduce platform to mine the passenger booking logs, used the LDA algorithm to mine passenger’s travel intentions, and then calculated the potential value of the airline. The experimental results on the passenger name records in 2011 indicate that the similarity(Jacarrd Index) between the airlines excavated and the actual hot airlines is up to 92%, which is 2% higher than the conventional method based on the number of flights. It can predict the future value of the airline more accurately and effectively.