Profile-Free and Real-Time Task Recommendation in Mobile Crowdsensing

As a key research issue in mobile crowdsensing (MCS), recent studies on task recommendation have begun to focus recommending tasks participants according the learned participant preferences. The common drawbacks of these are that, one hand, factors affecting preferences predefined, which is not practical as influential quite complex and full map profiles needs be preexisted. On other they do consider how update dynamically. To overcome drawbacks, profile-free real-time method proposed this work. First, we apply systems MCS realize recommendations. Second, participant-task-location tensor constructed, based an improved factorization presented provide recommendations for at given location. Finally, design algorithm idea time lists real time. Based real-world trace data sets, extensive evaluations show that has obvious advantages over baselines terms accuracy cost.