A game-based framework for crowdsourced data labeling

摘要

Data labeling, which assigns data with multiple classes, is indispensable for many applications, such as machine learning and data integration. However, existing labeling solutions either incur expensive cost for large datasets or produce noisy results. This paper introduces a cost-effective labeling approach and focuses on the labeling rule generation problem that aims to generate high-quality rules to largely reduce the labeling cost while preserving quality. To address the problem, we first generate candidate rules and then devise a game-based crowdsourcing approach CrowdGame to select high-quality rules by considering coverage and accuracy. CrowdGame employs two groups of crowd workers: One group answers rule validation tasks (whether a rule is valid) to play a role of rule generator, while the other group answers tuple checking tasks (whether the label of a data tuple is correct) to play a role of rule refuter. We let the two groups play a two-player game: Rule generator identifies high-quality rules with large coverage, while rule refuter tries to refute its opponent rule generator by checking some tuples that provide enough evidence to reject rules with low accuracy. This paper studies the challenges in CrowdGame. The first is to balance the trade-off between coverage and accuracy. We define the loss of a rule by considering the two factors. The second is rule accuracy estimation. We utilize Bayesian estimation to combine both rule validation and tuple checking tasks. The third is to select crowdsourcing tasks to fulfill the game-based framework for minimizing the loss. We introduce a minimax strategy and develop efficient task selection algorithms. We also develop a hybrid crowd-machine method for effective label assignment under budget-constrained crowdsourcing settings. We conduct experiments on entity matching and relation extraction, and the results show that our method outperforms state-of-the-art solutions.