Abstract:Crowdsourcing offers a new way for mobilizing the collective intelligence and efforts of a large group of people to tackle tasks which cannot be efficiently performed by machines (e.g., video transcription, monitoring road conditions at specific locations) (Doan, Ramakrishnan, and Halevy 2011). Many commercial crowdsourcing platforms are now available. Their business depends on providing satisfactory services to both the crowdsourcers (i.e. task requesters) and the workers involved. From the crowdsourcers' perspective, they expect to receive high quality results for their crowd sourcing tasks in a timely manner. From the workers' perspective, they want to earn as much as possible while committing limited productive effort.As the crowdsourcers are the main source of revenue for crowdsourcing platforms, their requirements tend to take precedence over those of the workers. Many commercial crowdsourcing platforms have implemented some variants of reputation-based mechanisms (Yu et al. 2013a) to gauge the workers' competence based on their past performance, and allow only those with good reputations to access tasks.While this simple reputation-based task delegation method is intuitive and has its own merits, a different, albeit related problem, has not received much attention. As crowd sourcing workers are human beings, they have limited availability and productive capacities to work on tasks delegated to them (Yu et al. 2012). Concentrating requests to workers with good reputations may result in details. In addition, as a small portion of reputable workers become overloaded with task requests while others remain relatively idle, attrition may occur in the worker population, thereby leaving the crowdsourcing platform with a shortage of workers.The potentially conflicting objectives between crowdsourcers and workers can be formalized under the congestion game framework (Monderer and Shapley 1996). In congestion games, the payoff for each player depends on the resources it selects and the number of other players selecting the same resources. For instance, the morning commute to work places by many people can be modeled as a congestion game. The time taken by a traveller on a given day depends on how many others are taking the same route as him.

Crowdsourcing Upon Learning: Energy-Aware Dispatch With Guarantee for Video Analytics

Trading off Coverage and Emergency for Hybrid Task Scheduling in Traffic Anomaly Detection

Crowd<sup>2</sup>: Multi-agent Bandit-based Dispatch for Video Analytics upon Crowdsourcing

Online Trichromatic Pickup and Delivery Scheduling in Spatial Crowdsourcing

MACRO: Incentivizing Multi-Leader Game-Based Pareto-Efficient Crowdsourcing for Video Analytics

Collaborative Task Allocation in Mobile Crowd Sensing.

A Multi-Objective Crowdsourcing Method for Mobile Video Streaming

Crowd Foraging: A QoS-Oriented Self-Organized Mobile Crowdsourcing Framework over Opportunistic Networks

Optimizing Worker Selection in Collaborative Mobile Crowdsourcing

Unknown Worker Recruitment with Budget and Covering Constraints for Mobile Crowdsensing

Scalable Urban Mobile Crowdsourcing

High Quality Participant Recruitment of Mobile Crowdsourcing over Big Data

Location-Aware Crowdsensing: Dynamic Task Assignment and Truth Inference

Optimizing Crowdsourcing Task Assignment Policies Using Multi-Agent Reinforcement Learning in Stochastic Games

Optimizing Cloud-Based Video Crowdsensing.

ActiveCrowd: A Framework for Optimized Multi-Task Allocation in Mobile Crowdsensing Systems

Infusing Human Factors Into Algorithmic Crowdsourcing

ActiveCrowd: A Framework for Optimized Multitask Allocation in Mobile Crowdsensing Systems

Toward QoI and Energy Efficiency in Participatory Crowdsourcing

Optimal Collaborative Uploading in Crowdsensing with Graph Learning

Budgeted Unknown Worker Recruitment for Heterogeneous Crowdsensing Using CMAB