The rapid growth of the airline industry has resulted in the availability of a large number of flights, however this can also create a paralyzing problem. Flight information on all airlines across the world can be obtained via the Internet. Today, passengers trend to be interested in user personalized service. How to effectively find a passenger’s most preferred air travel plan, which might include multiple transfers from millions of possible choices with certain constraints, such as time and price, is a critical challenge. This paper presents an efficient air travel planning approach, which can find a number of air travel plans by invoking the APIs offered by airline companies. At the same time, these plans also best match the customer’s preference based on an analysis of historical orders. An algorithm to extract user preference features is introduced and heuristic rules to speed up the K path search process under constraints are presented. The experiment results show that the proposed model finds optimal air travel plans efficiently on a real-world dataset.

中文翻译：

---

在云上飞行之前在 API 云上进行规划：一种实时个性化的航空旅行规划方法

航空业的快速增长导致了大量航班的可用性，但这也可能造成瘫痪问题。全世界所有航空公司的航班信息都可以通过互联网获得。今天，乘客趋向于对用户个性化服务感兴趣。如何有效地找到乘客最喜欢的航空旅行计划，其中可能包括从数百万种可能的选择中进行多次中转，并具有一定的限制，例如时间和价格，是一个关键的挑战。本文提出了一种高效的航空旅行计划方法，它可以通过调用航空公司提供的 API 来查找多个航空旅行计划。同时，根据对历史订单的分析，这些计划也最符合客户的偏好。介绍了一种提取用户偏好特征的算法，并提出了在约束条件下加速K路径搜索过程的启发式规则。实验结果表明，所提出的模型可以在现实世界的数据集上有效地找到最佳的航空旅行计划。