Publication Type

Conference Proceeding Article

Version

acceptedVersion

Publication Date

4-2020

Abstract

With the rapid development of mobile networks and the widespread usage of mobile devices, spatial crowdsourcing, which refers to assigning location-based tasks to moving workers, has drawn increasing attention. One of the major issues in spatial crowdsourcing is task assignment, which allocates tasks to appropriate workers. However, existing works generally assume the static offline scenarios, where the spatio-temporal information of all the workers and tasks is determined and known a priori. Ignorance of the dynamic spatio-temporal distributions of workers and tasks can often lead to poor assignment results. In this work we study a novel spatial crowdsourcing problem, namely Predictive Task Assignment (PTA), which aims to maximize the number of assigned tasks by taking into account both current and future workers/tasks that enter the system dynamically with location unknown in advance. We propose a two-phase data-driven framework. The prediction phase hybrids different learning models to predict the locations and routes of future workers and designs a graph embedding approach to estimate the distribution of future tasks. In the assignment component, we propose both greedy algorithm for large-scale applications and optimal algorithm with graph partition based decomposition. Extensive experiments on two real datasets demonstrate the effectiveness of our framework.

Keywords

prediction, task assignment, spatial crowdsourcing

Discipline

Databases and Information Systems | Data Science

Research Areas

Data Science and Engineering

Publication

2020 36th IEEE International Conference on Data Engineering (ICDE): Dallas, Texas; April 20-24: Proceedings

First Page

13

Last Page

24

ISBN

9781728129037

Identifier

10.1109/ICDE48307.2020.00009

Publisher

IEEE Computer Society

City or Country

Los Alamitos, CA

Copyright Owner and License

Authors

Additional URL

https://doi.org/10.1109/ICDE48307.2020.00009

Share

COinS