Building a task blacklist for online social platforms

Abstract

Recently, the use of crowdsourcing platforms (e.g., Amazon Mechanical Turk) has boomed because of their flexible and cost-effective nature, which benefits both requestors and workers. However, some requestors misused power of the crowdsourcing platforms by creating malicious tasks, which targeted manipulating search results, leaving fake reviews, etc. Crowdsourced manipulation reduces the quality of online social media, and threatens the social values and security of the cyberspace as a whole. To help solve this problem, we build a classification model which filters out malicious campaigns from a large number of campaigns crawled from several popular crowdsourcing platforms. We then build a task blacklist web service, which provides users with a keyword-based search so that they can understand, moderate and eliminate potential malicious campaigns from the Web.