Context-aware result inference in crowdsourcing

摘要: Abstract Many result inference methods have been proposed to address the quality-control problem in crowdsourcing. However, existing are ineffective for context-sensitive tasks ( CSTs ), e.g., handwriting recognition, translation, speech transcription, where context correlation within a task cannot be ignored two reasons. Firstly, it is crowdsource whole CST (e.g., recognizing handwritten texts) and use task-level infer answer, because rather hard correctly complete complicated task. Secondly, although composed of set atomic subtasks word), unsuitable split into multiple adopt subtask-level algorithm result, this will lose phrases) among increase difficulty Thus calls new approach handling . In work, we study propose context-aware algorithm. We design an by incorporating information. Furthermore, introduce iterative method improve quality. The results experiments on real-world demonstrated superiority our compared with state-of-the-art methods.