We study the following multi-period multi-objective online ride-matching problem. A ride-sourcing platform needs to match passengers and drivers in real time without observing future information, considering multiple objectives such as platform revenue, pick-up distance, and service quality. We develop an efficient online matching policy that adaptively balances the trade-offs among multiple objectives in a dynamic setting, and provide theoretical performance guarantee for the policy. We prove that the proposed adaptive matching policy can achieve a “target-based optimal solution”, i.e., a solution that minimizes the Euclidean distance to any pre-determined multi-objective target. Specifically, the outcome under our policy converges to the “compromise solution” if we set the utopia point as the target. Through numerical experiments and industrial testing using real data from a ride-sourcing platform, we demonstrate that our approach indeed obtains solutions that are closest to the pre-determined targets under various settings, in comparison to existing approaches. The policy presents solutions with delicate balance among multiple objectives and brings value to all the stakeholders in the ride-sourcing ecosystem comparing to benchmark policies: (1) drivers with higher service scores are dispatched with more orders and receive higher incomes; (2) passengers are more likely to be served by drivers with higher service scores, and passengers with higher order revenues are served with higher answer rates, at the expense of a small increase in pick-up distance; (3) the platform obtains a higher total revenue.
Menu
