A Long-Term Quality Perception Incentive Strategy for Crowdsourcing Environments with Budget Constraints
Liping Gao, Tao Jin, Chao Lü · International Journal of Cooperative Information Systems · 2020
Quality control is a critical design goal for crowdsourcing. However, when measuring the long-term quality of workers, the existing strategies do not make effective use of workers’ historical information, whereas others regard workers’ conditions as fixed values, even if they do not consider the impact of workers’ quality. This paper proposes a long-term quality perception incentive model (called QAI model) in a crowdsourcing environment with budget constraints. In this work, QAI divides the entire long-term activity cycle into multiple stages based on proportional allocation rules. Each stage treats the interaction between the requester and the worker as a reverse auction process. At each stage, a truthful, individually rational, budget feasible, quality-aware task allocation algorithm is designed. At the end of each stage, according to hidden Markov model (HMM), this paper proposes a new framework for quality prediction and parameter learning framework, which can make use of workers’ historical information efficiently. Experiments have verified the feasibility of our algorithm and showed that the proposed QAI model leads to improved results.