[1J2-J-6] Web mining 1

[1J2-J-6-04] Proposal of Flexible Route Recommendation Method Employing Edge Vector-Based Formulation

〇Hiroki Shibata¹, Yasufumi Takama¹ (1. Tokyo Metropolitan University)

Keywords:Sightseeing Route Recommendation, Recommendation, Simulated Annealing, Optimization

Many studies for sightseeing route recommendation use the formulation of Selective Traveling Salesman Problem or its extension. However, this formulation uses the costs of the shortest paths between spots as the weights of edges in a graph, which means that the variety of routes between spots is ignored. Uses might prefer different routes than the shortest path, such as the route along the river, forest, or sea if they have enough time. This paper proposes a method to recommend routes by considering the balance between staying time and traveling time. Extending the Edge Vector-based formulation, the proposed method utilizes inner routes in a spot to represent the staying time as the costs of those inner routes. An experiment using an artificial data set shows that the proposed method can optimize the staying time and traveling time concurrently.