DOCS

摘要: Crowdsourcing is a new computing paradigm that harnesses human effort to solve computer-hard problems, such as entity resolution and photo tagging. The crowd (or workers) have diverse qualities it important effectively model worker's quality. Most of existing worker models assume workers the same quality on different tasks. In practice, however, tasks belong variety domains, domains. For example, who basketball fan should better for task labeling related 'Stephen Curry' than one 'Leonardo DiCaprio'. this paper, we study how leverage domain knowledge accurately We examine using base (KB), e.g., Wikipedia Freebase, detect domains workers. develop Domain Vector Estimation, which analyzes with respect KB. also Truth Inference, utilizes domain-sensitive infer true answer task. design an Online Task Assignment algorithm, judiciously efficiently assigns appropriate To implement these solutions, built DOCS, system deployed Amazon Mechanical Turk. Experiments show DOCS performs much state-of-the-art approaches.