Nipping trouble in the bud: A proactive tourism recommender system

Abstract

Recommender systems have been used extensively in tourism activities to aid travel decision making. However, existing systems design routes using mainly deterministic information, ignoring the negative impacts of uncertain events, likely resulting in suboptimal or impractical route designs. This study addresses this issue by integrating forecasting mechanisms into system design, thereby enabling tourists to avoid potential risks and improving travel experience. For example, in crowding avoidance, the system's performance is demonstrated through a field experiment. The results suggest that our system outperforms benchmarks and can help tourists avoid attraction crowding. The system is also applicable for other predictable uncertain events.