A personalized recommendation system for multi-modal transportation systems

Recommendation system has recently experienced widespread applications in fields like advertising and streaming platforms. Its ability of extracting valuable information from complex data makes it a promising tool for multi-modal transportation system. In this paper, we propose a conceptual framework for proactive travel mode recommendation combining recommendation system and transportation engineering. The proposed framework works by learning from historical user behavioral preferences and ranking the candidate travel modes. In this framework, an incremental scanning method with multiple time windows is designed to acquire multi-scale features from user behaviors. In addition, to alleviate the computational burden brought by the large data size, a hierarchical behavior structure is developed. To further allow for social benefits, the proposed framework proposes to adjust the candidate modes according to real-time traffic states, which is potential in promoting the use of public transport, alleviating traffic congestion, and reducing environmental pollution.