Research Of Online Task Allocation Strategy In Spatio-Temporal Crowdsourcing Environment

With the development of mobile Internet and O2 O business model as well as the popularity of intelligent mobile devices,spatio-temporal crowdsourcing platforms such as Didi Taxi,ele.me and other applications have become more and more popular and are closely related to human life.Spatio-temporal crowdsourcing platforms has high requirements on real-time response and rationality of task allocation.The quality of task allocation will directly affect the performance of crowdsourcing platforms.Based on the above reasons,task allocation under the spatio-temporal crowdsourcing environment becomes an important issue.Therefore,this paper has practical significance for the research of task allocation under the environment of spatio-temporal crowdsourcing.Task allocation is one of the core research issues among the spatio-temporal crowdsourcing platforms,it aims to match randomly generated crowdsourcing tasks,crowdsourcing workers and workplaces,and get the best utility.In the process of task allocation,the threshold algorithm plays a key role in filtering low-utility allocation,which often determines the utility of entire crowdsourcing platform.In previous studies,greedy algorithm,the hypothesis random threshold algorithm and adaptive threshold algorithm couldn't be adapt to the real crowdsourcing environment,and there exists the following problems: focusing on the setting of threshold,although the setting of threshold value can filter out the allocation with less utility,there is still some randomness in the task allocation process,which often results in less than expected results,the existed task allocation strategy needs to be optimized.The research contents of this paper are as follows: First of all,the crowdsourcing is briefly introduced,and the research results at domestic and overseas are expounded,focusing on online task allocation strategy namely threshold algorithm and the region division algorithm under the environment of spatio-temporal crowdsourcing.Based on the analysis of the action mechanism of threshold algorithm,it is found that the problem of low utility will appear when data are not uniform,an adaptive threshold algorithm based on statistical prediction and matching strategy is proposed,effectively reduce the waste utility.According to the characteristics of the location of the crowdsourcing task and task distribution,the ARD region division algorithm is proposed,it can filter low-return value tasks more accurately and improve the utility value obtained.Finally,the experiment platform is built and the algorithms were verified with real data.Experimental results show that the proposed algorithms can effectively improve the utility value of the crowdsourcing platforms.