An integrated case-based reasoning approach for personalized itinerary search in multimodal transportation systems

• A new approach for personalized itinerary search in the field of multi-modal transport is developed.
• CBR and the Choquet integral are integrated to suggest the best itinerary based on user’s preferences.
• User’s behavior is predicted by comparing his preferences to those of the past users.
• This method provides enhanced accuracy and speed in case matching along with computational efficiency.
• This approach can facilitate intelligent decision support in the field of transport.

Suggesting personalized itinerary search for travelers in a multimodal transportation system is a challenging problem. This is due to the increased complexity and diversity of transportation means, the intricacy and multitude of destinations along with the amount of rapidly changing information available to the traveler. Providing the transportation user with the relevant information that only meets his needs, preferences and personal profile is of foremost importance in efficiently supporting passenger mobility requirements in a large urban agglomeration.In this paper, we propose a multi-criteria approach for suggesting personalized itinerary to transportation users based on their preferences and needs. The proposed approach integrates case-based reasoning with Choquet integral to suggest the itinerary that best matches the user’s preferences. Further, the proposed method predicts the user’s behavior by comparing his preferences to those of other users with the same preferences for a given context. This will help the user to adopt the best action when facing a new situation in his itinerary search.This will help the user adopt the best action facing a new situation. Personalized information retrieval is processed based on criteria which weights are determined using the two-additive Choquet integral. The performance of the proposed algorithm was assessed by solving a real-life itinerary planning problem defined in the Tunisian urban public transit network. A comparison study involving both qualitative and quantitative assessment of the proposed approach as compared to two other methods was also carried out. Based on the performance analysis, as well as the comparison study, our new approach provides the best solutions for applications requiring personalization based user’s preferences in a multi-criteria setting.