A Novel Crowdsourcing Task Recommendation Method Considering Workers’ Fuzzy Expectations: A Case of ZBJ.COM

Abstract

Crowdsourcing recommendation is necessary to surmount information overload and assist workers to identify tasks of interest efficiently. However, existing studies mainly concentrate on investigating workers’ qualifications or requirements, neglecting their fuzzy expectations and psychological behavior. Besides, workers can only passively accept the results and do not have any right to change them even if they are not satisfactory. As workers will pay different attention to task aspects, we regard the crowdsourcing recommendation as a multi-criteria decision-making (MCDM) problem. In this paper, we propose an MCDM-based method for automated and interactive crowdsourcing task recommendations while comprehensively considering workers’ fuzzy expectations and psychological behavior. Given a large volume of available tasks on a crowdsourcing platform, the requirement of massive human efforts in traditional MCDM is not applicable in such situation. Therefore, we formulate a set of quantitative task evaluation criteria for automated task assessments based on the attributes that have the most impact on workers’ participation behavior. Further, to capture workers’ fuzziness in expectations, an intuitionistic fuzzy 2-tuple linguistic (IF2L) term is introduced, which are further leveraged as reference points to evaluate their “gains” and “losses” based on the prospect theory. Finally, the D-S evidence theory is adopted to gather workers’ uncertain prospect values. A case of ZBJ.COM, a popular crowdsourcing platform in China, shows that the proposed method is efficient and effective in recommending suitable and personalized tasks to workers.