开放模式下群智感知高质量数据采集方法研究

移动群智感知利用人们携带的智能设备作为感知单元完成大规模的数据采集,是一种分布式和弱协作式的数据采集方式,因此,参与者常常有意或无意的采集并提交了错误或者重复的低质数据.在仅支付高质量数据的约束下,采用传统封闭采集模式时,参与者只能按照任务要求自行调整采集策略,为此,我们采用开放式采集模式,即向参与者提供已收集到的数据信息,帮助参与者调整采集策略,提高数据采集的质量和效率.我们招募到99名志愿者参加了22轮数据采集任务.为了评估开放式采集在按质支付方式下的效果,我们采用了传统支付和逆向拍卖两种激励机制,对比了不同数据采集方式、不同任务难度和不同支付方式下,数据质量和数据收集成本的变化.实验结果显示采用开放式采集可以有效降低数据冗余率,提高困难任务的数据质量,并且参与者采用逆向拍卖时的出价更合理,拍卖成功率也更高.

Research on the Method to Collect High-quality Crowdsourced Data in Open Mode

Mobile crowdsensing(MCS)treats people’s smart devices as sensing units to accomplish different large-scale data-collecting tasks.The wrong or duplicated crowdsourced data has low-quality.MCS is a distributed and weakly-cooperative data collection manner,thus,participants often intentionally or unintentionally get and submit low-quality data.Under the constraint of paying for only high-quality data,if the data is collected in traditional close mode,participants can only adjust the data-collecting strategy according to task requirements.For this reason,in order to help participants adjust their data-collecting strategies and improve the quality and efficiency of collecting data,we adopt the open data-collecting mode,i.e.,we provide participants the information of data that have uploaded to the cloud.We recruited 99 volunteers who participated in 22 data-collection tasks.For the purpose of evaluating the effectiveness of data-collecting in open mode under the pay-by-quality constraint,we used both the traditional payment and the reverse auction payment.In the experiment,we set up tasks with different difficulties,used different data-collecting modes and different payment methods,and then observed the changes of data quality and data collection cost.Experimental results showthat after in open-collecting mode,the data redundancy decreases and the quality of data for difficult tasks is improved,and participants bid their data more reasonably and win more in reverse auctions.