Portal de Periódicos da CAPES

Predicting Crowdsourcing Worker Performance with Knowledge Tracing

2020; Springer Science+Business Media; Linguagem: Inglês

10.1007/978-3-030-55393-7_32

1611-3349

Zizhe Wang, Hailong Sun, Tao Han,

Intelligent Tutoring Systems and Adaptive Learning

Knowledge-intensive crowdsourcing (KI-C) plays an important role in today's knowledge economy. And competitive knowledge-intensive crowdsourcing (CKI-C) is a kind of KI-C in which tasks are released in the form of competitions. The worker performance prediction is important for CKI-C platforms to recommend tasks to proper workers. Traditional worker performance prediction methods do not consider the complex properties of tasks and worker skills, thus they do not function in CKI-C. In this work, we design the KT4Crowd framework to incorporate knowledge tracing, used effectively in intelligent tutoring systems (ITS), into CKI-C for predicting worker performance. The experimental results on a large-scale Topcoder dataset show the effectiveness of our framework and the DKVMN model with our framework achieves the best performance among the compared state-of-the-art methods.