民航旅客座位偏好建模与应用研究

发布时间：2019-06-15 20:41

【摘要】：随着互联网与电子商务的飞速发展,如何准确地把握用户的偏好和需求,越来越成为电子商务领域的一个热点研究问题。准确的用户画像(User Profiling)和高效的用户偏好建模,在给用户带来精准的个性化服务与推荐的同时,也改善了电子商务企业的客户关系管理(CRM),提升了企业效益。同样地,在民航领域也存在着旅客偏好建模问题。在经济全球化、航空联盟化的趋势下,民航客运面临着激烈的竞争,民航企业千方百计地提高运营效率,从而提高自身行业竞争力。对于民航企业来说,挖掘出旅客对座位的偏好,提升对旅客的理解,给旅客提供针对性地个性化服务成为客户关系管理(CRM)的重要举措之一。尤其是,根据旅客的偏好给航班上的旅客分配座位、提升绝大部分旅客的出行满意度,成为航空公司竞相追求的目标。本文从社会网络的视角研究了民航旅客座位偏好建模问题,综合考虑民航领域座位资源的有限性和旅客间的社会关系,提出了一种旅客个体偏好和旅客关系偏好相结合的建模方法。首先,我们利用民航旅客历史出行记录推断出旅客间的社会关系,并利用这些关系构建旅客共同出行网络。接着,基于座位资源的有限性,构建旅客个体对座位属性的偏好模型。然后,基于旅客共同出行网络,探索旅客关系间的亲密程度与座位距离的相关性,构建旅客关系偏好模型。接着,我们采用线性组合的方式将旅客个体偏好和旅客关系偏好结合起来。最后,基于旅客偏好模型给旅客推荐座位,并采用先来先服务算法给整个航班旅客分配座位。在民航领域的一个真实的数据集上进行了实验,证明本文提出的偏好模型能够有机地将旅客个体偏好和关系偏好结合起来,较好地描述旅客对座位的偏好,为旅客座位推荐问题提供了理论依据。此外,实验还表明与现实中旅客值机过程生成的座位情况相比,本文采用先来先服务算法进行座位分配提高了旅客的满意度。
[Abstract]:With the rapid development of Internet and electronic commerce, how to accurately grasp the preferences and needs of users has become a hot research issue in the field of electronic commerce. Accurate user portrait (User Profiling) and efficient user preference modeling not only bring accurate personalized service and recommendation to users, but also improve the customer relationship management (CRM), of e-commerce enterprises to improve the efficiency of the enterprise. Similarly, there is a problem of passenger preference modeling in the field of civil aviation. Under the trend of economic globalization and aviation alliance, civil aviation passenger transport is facing fierce competition. Civil aviation enterprises do everything possible to improve their operational efficiency, so as to improve the competitiveness of their own industries. For civil aviation enterprises, excavating passengers' preference for seats, improving their understanding of passengers, and providing targeted personalized service to passengers has become one of the important measures of customer relationship management (CRM). In particular, according to the preferences of passengers to allocate seats to the passengers on the flight, to improve the travel satisfaction of the vast majority of passengers, has become the goal of airlines competing to pursue. In this paper, the modeling problem of civil aviation passenger seat preference is studied from the perspective of social network. Considering the limitation of seat resources in the field of civil aviation and the social relations between passengers, a modeling method combining passenger individual preference and passenger relationship preference is proposed. First of all, we use the historical travel records of civil aviation passengers to infer the social relations between passengers, and use these relations to construct the common travel network of passengers. Then, based on the limitation of seat resources, the preference model of passengers to seat attributes is constructed. Then, based on the common travel network of passengers, the correlation between the intimacy of passenger relationship and seat distance is explored, and the preference model of passenger relationship is constructed. Then, we combine passenger individual preference with passenger relationship preference by linear combination. Finally, based on the passenger preference model, passengers are recommended for seats, and the first-come-first-served algorithm is used to assign seats to the whole flight. An experiment is carried out on a real data set in the field of civil aviation. It is proved that the preference model proposed in this paper can organically combine the individual preference and relational preference of passengers to describe the preference of passengers to seats, which provides a theoretical basis for the problem of passenger seat recommendation. In addition, the experiment also shows that compared with the seat situation generated by the passenger check-in process in reality, this paper uses the first-come-first-served algorithm to allocate seats to improve the satisfaction of passengers.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.9

【共引文献】