부산대학교 도서관

Ridesharing has the potential to contribute to reducing CO2 emissions and congestions but user acceptance is still low. Within ridesharing platforms, the assignment of ride offers to requests and vice versa is an essential process with respect to user satisfaction and user acceptance. Our research aims to contribute to increase ridesharing usage by improving this process. In particular, we strive to enhance existing assignment approaches that are based on deep reinforcement learning by including additional factors that influence the satisfaction of users. Furthermore, we present the novel concept of applying explainable AI techniques to make ridesharing assignment decisions more transparent to users. In this paper, we present a five-step research methodology to realize the aforementioned enhancements in the future and present preliminary results.