Task Selection Based on Worker Performance Prediction in Gamified Crowdsourcing

This paper presents a method for selecting tasks in gamified crowdsourcing. Knowledge collection is important for constructing a high-quality knowledge base. The knowledge collection process is targeted in our word retrieval assistant system, in which knowledge is presented as triples. Four types of quizzes are introduced that can be used to collect knowledge from many casual users. The quizzes are variations of a fill-in-the-blank format, in which the user provides a piece of information by filling in the blanks in the quiz. To collect knowledge efficiently even when the required knowledge is distributed among many users, a prediction method is introduced to select a quiz that is best suited for a particular user. The simulation results demonstrate the potential of the proposed method.