TOP-Key Influential Nodes for Opinion Leaders Identification in Travel Recommender Systems.

Travel recommender systems, also called (TRS) have recently gained significant attention in the research and industrial communities. These systems aim at identifying the travellers preferences and providing adequate suggestions to them whenever and wherever they want. Thus, TRS are very helpful for travelers, particularly, when they visit a place they have never been to before. Opinion Leaders based-technique attempts to identify the set of most important delegates who can represent as many TRS users as possible to alleviate the cold start user problem when a new user is registered to the system and has no ratings yet and the cold start item problem when a new item is added to the system and has no interactions yet. In this paper, we propose two graph based approaches for Opinion Leaders Detection in Travel Recommender Systems. The first is based on the Minimum Cover Vertex identification, while the second uses the Fragmentation method to detect the set of most influential nodes in the recommender system graph. The obtained experimental results confirm the effectiveness of our proposal.