Towards Fairness Issues in Spatial Crowdsourcing

PhD Thesis Proposal Defence


Title: "Towards Fairness Issues in Spatial Crowdsourcing"

by

Mr. Zhao CHEN


Abstract:

With the booming of mobile Internet and sharing economy, online spatial 
crowdsourcing services, such as Uber and Didi, are becoming important 
infrastructures of our daily life. Existing studies about spatial 
crowdsourcing usually focus on the platform interests and ignore 
experiences of requesters and workers. Because of the dynamic task demand 
and worker supply changes, the major experience issue comes from 
insufficient valid assignments for both workers and requesters. In this 
thesis we discuss the problem of how to allocate limited assignment 
resource fairly to both worker and requester sides.

For a requester, the resource is available nearby workers because his/her 
tasks need to be assigned as soon as possible. Thus, we study the 
minimizing maximum delay spatial crowdsourcing (MMD-SC) problem and 
propose solutions aiming at achieving a worst case controlled task 
assignment. The MMD-SC problem assumes that both workers and requesters 
come dynamically  and considers not only the workers' travel costs but 
also the buffering time of tasks, thus it is very challenging due to 
two-sided online setting. To address these challenges, we propose a space 
embedding based online random algorithm  and two efficient heuristic 
algorithms.

For the worker side, the resource is new tasks which needs to be 
distributed to workers as equal as possible.The first challenge of this 
problem is how to formally define the fairness measurement which can 
indicate whether it is fair enough for workers in a many-to-many 
task-and-worker matching and the second one is how to do online 
assignment/matching accordingly. To address these challenges, we first 
propose a novel definition of fairness based on the worker-share in the 
assignment and show its validity under several fairness principles in 
existing fairness scheduling studies. Next, we formally define the 
worker-fairness-aware assignment problem (WFAA) and propose a min-max 
matching based assignment algorithm accordingly.

In addition, we show the effectiveness and efficiency of our methods via 
extensive experiments on both synthetic and real datasets.


Date:			Monday, 9 December 2019

Time:                  	4:00pm - 6:00pm

Venue:                  Room 2132C
                         (lift 19)

Committee Members:	Prof. Lei Chen (Supervisor)
 			Dr. Raymond Wong (Chairperson)
 			Dr. Xiaojuan Ma
 			Dr. Wei Wang


**** ALL are Welcome ****