Passenger Prediction in Shared Accounts for Flight Service Recommendation

• Published in: Advances in Services Computing Vol. 10065; pp. 159 - 172
• Main Authors: Zhao, Yafeng, Cao, Jian, Tan, Yudong
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2016; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Dirichlet distribution; Gibbs sampling; Information retrieval; Probabilistic topic model; Recommender system; Shared account
• ISBN: 3319491776; 9783319491776
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-49178-3_12

Personalized recommendation is needed for online flight booking service because it is a difficult task for a traveller to select the flight when the number of available flights is large. Traditionally, we can recommend flights to a user based on his historical orders collected from his account. However, people sometimes book tickets for his family members, friends or colleagues through his account. In this case, the preferences of other travellers should also be considered. Unfortunately, before placing the order, people will not provide passengers’ information. Therefore, we propose a probabilistic method for passenger prediction based on historical behaviors and contextual knowledge. We then experimentally demonstrate its effectiveness on a real dataset. The result shows that our method outperforms conventional methods.