An approach to task recommendation in crowdsourcing based on 2-tuple fuzzy linguistic method

Abstract

Purpose

Task recommendation is an important way for workers and requesters to get better outcomes in shorter time in crowdsourcing. This paper aims to propose an approach based on 2-tuple fuzzy linguistic method to recommend tasks to the workers who would be capable of completing and accept them.

Design/methodology/approach

In this paper, worker’s capability-to-complete (CTC) and possibility-to-accept (PTA) for a task needs to be recommended are proposed, measured and aggregated to determine worker’s priority for task recommendation. Therein, the similarity between the recommended task and its similar tasks and worker’s performance on these similar tasks are computed and aggregated to determine worker’s CTC quantitatively. In addition, two factors of worker’s active degree and worker’s preferences to a task category are presented to reflect and determine worker’s PTA. In the process of measuring them, 2-tuple fuzzy linguistic method is used to represent, process and aggregate vague and imprecise information.

Findings

To demonstrate the implementation process and performance of the proposed approach, an illustrative example is conducted on Taskcn, a widely used Chinese online crowdsourcing market. The experimental results show that the proposed approach outperformed the self-selection approach, especially for complex or creative tasks. Moreover, comparing with task recommendation considering worker’s CTC solely, the proposed approach would be better in terms of workers’ response rate. Additionally, the use of linguistic terms and fuzzy linguistic method facilitates the expression of vague and subjective information and makes recommendation process more practical.

Research limitations/implications

In the study, the authors capture alternative workers, collect workers’ behaviors and compute workers’ CTC and PTA manually. However, as the number of tasks and alternative workers grow, the issue, i.e. how to conveniently collect workers’ behaviors and determine their CTC and PTA, becomes conspicuous and needs to be studied further.

Practical implications

The proposed approach provides an alternative way to perform tasks posted in crowdsourcing platforms. It can assist workers to contribute to right tasks, and requesters to get outcomes with high quality more efficiently.

Originality/value

This study proposes an approach to task recommendation in crowdsourcing that integrates workers’ CTC and PTA for the recommended tasks and can deal with vague and imprecise information.

Citation

Zhang, X. and Su, J. (2018), "An approach to task recommendation in crowdsourcing based on 2-tuple fuzzy linguistic method", Kybernetes, Vol. 47 No. 8, pp. 1623-1641. https://doi.org/10.1108/K-12-2017-0468