Game Models And Mechanism Design In Sharing Economy

With the improvement and development of various mobile technology,sharing economy has achieved great success in practice.Applications such as ride sharing,bike sharing and charger sharing banks,have saved resources and brought great convenience to society and individuals.At the same time,sharing economy is new economy model and there are many new problems to be studied.Sharing economy is different from the traditional market.Ride sharing platforms such as Didi,Uber and all kinds of sharing economy platforms do not directly manage demand or supply resource.They usually play the role of resource scheduling and pricing.Platforms need to coordinate supply and demand to pursue maximization of the total system benefits.The two-sided market economy model applied is with many problems to be studied.Take the most mature ride-sharing system as an example,participation of drivers and passengers is completely voluntary and bilateral participation in the market has great influence on the good operation of the entire market and the total benefit of the platform.Therefore,how to design an effective incentive mechanism to attract participation of passengers/drivers is an important and critical issue.In ride-sharing systems,due to the complexity of cost functions,classical models and strategies of traditional game theory and mechanism design are not applicable.New models and algorithm designs are needed.This article mainly focus on incentive mechanism design and analysis in ride sharing system and bike sharing system.The designs aim at making the best use of resource and pursuing the maximal benefits of the whole system.The efficiency of the designs is proved in theory and by evaluation experiments.There are four following parts in this article:First,we model ridesharing problem with complex traveling costs in game theory of variable costs which has unique advantage in resource allocation.Under this model,we design efficient mechanism satisfying budget balance.It is different from the traditional game theory where the cost of items are usually set as zero or independent of allocation result.We want truthful mechanisms with desirable properties.We prove that no truthful mechanism with desirable properties can get better than 2-approximation in social welfare in a variable reserve price auction model.We design mechanism with the best approximation ratio in theory and test it on real dataset.The evaluation test shows that although the approximation in theory is 2,the evaluation result is much better due to data property.What’s more,though revenue is not the objective of our design,our mechanism performs well on it.We applied such model and our mechanism in bike sharing system to solve the incentive problem of bike fixed-point parking and get good results in theory and evaluation.Second,in online system,demand of passengers could arrive at any time and the system must response in a short time.The platform need to make decisions without full information which is different from the traditional offline setting.We need to deal with the challenge of online mechanism design.We design online mechanisms for ride sharing system to maximize the number of passengers taken and maximize the total system benefits respectively.For different objectives,we get different results.We cannot get any approximation guarantee for social welfare maximization but rather good approximation guarantee for other two objectives.We analysis our designs in theory and evaluate them on real dataset.All of them perform well.Third,there are several passengers share one car in practice which leads to complex routing.Such scenario makes it difficult share cost/payment and optimize the whole system objective.What’s worse,the fairness should also be considered.It is difficult to handle such problems and we clarify that the traditional mechanisms are not appropriate.We design cost sharing mechanism with good approximation guarantee for such situation.Fourth,in the sharing economy,many participants share resources and the influence among participants are unavoidable.This can be modeled as game with externalities,that is,the values agents get is influenced by the allocation results of other agents.For instance,in the situation where several passengers share one car,different passengers may prefer different seats and the allocation of seats will influence the value of the passengers.Such model is not fully studied and we start with a simple situation with negative externalities to design mechanisms with desirable properties.We design truthful,efficient mechanisms with polynomial time complexity for single competitor graph,planar graph and general graph.Inspired by such results,we design mechanisms for the game model with identity-dependent negative externalities and design algorithm for ridesharing system considering seat preference.