Crowdsourcing information systems aims at delivering informational products and services by harnessing a large group of online users. Individuals motivated intrinsically (e.g., enjoyment) or extrinsically (e.g., reward) can contribute to the system through picking among a variety of open tasks on crowdsourcing platforms. As the huge amount of tasks posted ever day, matching individual with an appropriate task that meet up individual’s personal preference and skill is crucial to the success. However, in reality, the ever-increasing amount of opportunities engaging individuals on crowdsourcing platforms lead to an information overload situation. Therefore, how to assist contributors in finding a suitable task in line with self-identification principle has attracted scholars and practitioners’ attentions. Geiger and Schader (2014) review and analyze the current state of personalized task recommendation in crowdsourcing context which shed a light on designing the relevant mechanisms on crowdsourcing platforms.
Similar to any other information systems, a classification based on the functionality is required for crowdsourcing system as well. In this article, the authors categorize crowdsourcing system based on two distinct dimensions. The first one is whether the system seeks homogeneous or heterogeneous contributions and the second one is whether the system seeks a non-emergent or an emergent value from the contributions. More specifically, contributions are equally valued when a system seeks homogeneous contribution as opposed to a system seeks heterogeneous contributions according to the quality of individual submissions. Individuals’ contributions are valued separately when a system focus on non-emergent value whereas the emergent value is derived from the collective of contributions. According to the combination of these two dimensions, the authors come up with four archetypes of crowdsourcing information systems along their functions, which is shown in the figure below:
The findings of this article are generated into two aspects. The first finding concentrates on the current development of recommendation approaches under different context of crowdsourcing information systems. For instance, implementing task assignment in contrast to the task recommendation to match suitable tasks to contributions are common in crowd processing systems. In crowd solving systems, scholars proposed a social search engine which assist in assigning tasks to individuals who have higher probability to solve them. The second finding indicates ways of generating better personalised recommendation system in crowdsourcing context. Firstly, by taking advantage of individual knowledge, systems are able to record both contributors’ implicit (e.g., timeframe of completing tasks) and explicit (e.g., opinions and comments of tasks) behaviors to assess their interests and capabilities. Secondly, through connecting external knowledge of contributors from other social networks or information systems, the recommendation systems can have more precise profiles of each contributors. Thirdly, the quality of recommendation systems can be improved by linking content-based and collaborative filtering techniques as well as additional information such as demographic and social relations.
Personalized task recommendation in crowdsourcing context has certain practical significance. From the perspective of contributors, personalized task recommendation helps potential contributors in search of suitable tasks which match their preferences and capabilities. Accurately involving into the right tasks can further result in higher motivation of contributors. From the perspective of requesters, personalized task recommendation is able to increase the chances of gathering contributors with right skills which allows requesters have higher probability to gain the best solution.
Geiger, D., & Schader, M. (2014). Personalized task recommendation in crowdsourcing information systems – Current state of the art. Decision Support Systems, 65(C), 3–16.