Abstract:

Current allocation methods of spatiotemporal crowdsourcing task are mostly targeted at crowdsourcing workers who have incentive condition constraints and full-time crowdsourcing task, ignoring interest-oriented crowdsourcing workers who have interest preferences and fulfil tasks without reward constraint. The problem how to recommend crowdsourcing tasks to these interest-oriented workers needs to be solved urgently. In view of this, this method considers the spatiotemporal behavior rules and interest preferences of spatiotemporal crowdsourcing workers. This method firstly introduces the gini coefficient and selects the data of interest-oriented spatiotemporal crowdsourcing workers from the data. The clustering method of the geo-social relationship model is used to cluster the crowdsourcing tasks. The Markov model based on Gaussian analysis is used to predict the probability of various locations that the crowdsourcing workers may arrive at the next time point. The tasks are recommended to the interest-oriented workers in descending order of probability. Experimental results show that the proposed method effectively improves the completion rate of interest-oriented spatiotemporal crowdsourcing tasks.

Key words: spatiotemporal crowdsoourcing technology, interest-oriented spatiotemporal crowdsourcing workers, spatiotemporal behavior analysis

摘要：

当前的时空众包任务推荐方法大都是针对有奖励约束、全职做众包任务的众包工人，忽略了有兴趣偏好、不受奖励约束完成任务的兴趣型众包工人，如何将众包任务推荐给这些兴趣型工人，是亟待解决的问题。针对此情况，提出考虑兴趣型时空众包工人的时空行为规律和兴趣偏好的推荐方法。引入基尼系数，在数据中筛选出兴趣型时空众包工人的数据，利用地理-社会关系模型的聚类方法对众包任务进行聚类，用高斯分析的马尔可夫模型预测众包工人在下一转移时间点可能到达各个地点的概率，把位于众包工人可能到达地点的任务按概率降序推荐给兴趣型工人。实验结果表明，所提方法有效提高了兴趣型时空众包任务的完成率。

关键词: 时空众包, 兴趣型时空众包工人, 时空行为分析