改进二部分图的民航旅客出行偏好模式的构建

CONSTRUCTING AVIATION PASSENGER TRAVEL PREFERENCES PATTERN WITH IMPROVED BIPARTITE GRAPH

根据旅客的个人信息、历史购买信息、常旅客信息以及旅客通过终端设备的历史访问记录,实现自动的旅客出行偏好模式的构建。根据原始的二部分图资源分配算法提出一种对资源分配矩阵的改进方法,并且与原始矩阵进行耦合,以提高推荐的精确度。实验结果表明该改进方法的精确度随着可变参数的变化而变化,当可变参数取某一最优化值时,改进后的方法精确度优于改进前的。将改进后的方法用于构建旅客出行偏好模式,应用后的结果与原始数据集比较,的确发现了旅客出行的规律。

According to passenger＇s personal information, purchase history information, frequent flyer information and historical access re- cords of guests passing the terminal equipments, the automatic construction of passenger travel preferences pattern is realised. According to raw bipartite graph resource allocation algorithm we propose an improved approach for resources allocation matrix, and couple it with the origi- nal matrix to improve recommendation accuracy. Experimental results show that the accuracy of this improved method varies along with the changes of variable parameters, when the variable parameter is assigned a certain optimisation value, the accuracy of the improved method is superior to the one not improved. Applying the improved method to constructing the preference pattern of passenger travel, it does find the passenger travel rules by comparing the applied results with the original data set.