Diversification of Route Planning Results for Improved User Satisfaction

While existing route planning techniques for public transit networks are efficient in finding optimal routes, we argue that users could be confused by these theoretical optimal yet impractical routes. To improve user satisfaction, we develop PARTNER - a practical route planning system that can provide users with options to take the next best route if they miss the optimal one. These options also include other preferences, e.g., fewer number of transfers, as long as they are reasonable in terms of total travel time. In addition, we consider a realistic public transit network covering a wide geographical area with sub-regions and different travel calendars, as well as location-to-location queries which require geocoding and reverse geocoding capabilities, and walking edges between nearby stops for transfers across public transit services. Experiments on a realistic transit network confirm the robustness of the proposed algorithm as well as the practicality of the returned routes.