Verbal vs. Nonverbal Cues in Static and Dynamic Contexts of Fraud Detection in Crowdsourcing

Abstract

As an important mode of open innovation, crowdsourcing can effectively integrate external resources, enabling enterprises to obtain stronger competitiveness and more benefits at a faster speed and lower cost. However, this mode has inevitable intellectual property protection challenges, especially on contest-based crowdsourcing platforms. Previous studies mostly focused on the protection of the rights of sponsors while ignoring the rights of workers, rarely paying attention to sponsor fraud, which may reduce the enthusiasm of participants and eventually turn crowdsourcing;' into a lemon market. This study proposes several fraud detection models to address this problem on contest-based crowdsourcing platforms. Furthermore, this paper explores and compares the value of four types of information as deception cues in crowdsourcing contexts via data mining technology and machine learning methods. The results benefit participants in crowdsourcing markets and contribute to fraud detection research and open innovation in the knowledge economy.