《Towards Social Norm Design for Crowdsourcing Markets 面向众包市场的社会规范设计》由会员分享，可在线阅读，更多相关《Towards Social Norm Design for Crowdsourcing Markets 面向众包市场的社会规范设计（7页珍藏版）》请在金锄头文库上搜索。
1、Towards Social Norm Design for Crowdsourcing MarketsChien-Ju Ho University of California, Los Angeles cjhocs.ucla.eduYu Zhang University of California, Los Angeles yuzhangucla.eduJennifer Wortman Vaughan University of California, Los Angeles jenncs.ucla.eduMihaela van der Schaar University of Califo
2、rnia, Los Angeles mihaelaee.ucla.eduAbstractCrowdsourcing markets, such as Amazon Mechanical Turk, provide a platform for matching prospective workers around the world with tasks. However, they are often plagued by workers who attempt to exert as little effort as possible, and requesters who deny wo
3、rkers payment for their labor. For crowdsourcing markets to succeed, it is essential to discour- age such behavior. With this in mind, we propose a frame- work for the design and analysis of incentive mechanisms based on social norms, which consist of a set of rules that participants are expected to
4、 follow, and a mechanism for up- dating participants public reputations based on whether or not they do. We start by considering the most basic version of our model, which contains only homogeneous participants andrandomlymatchesworkerswithtasks. Theoptimalsocial norm in this setting turns out to be
5、 a simple, easily compre- hensible incentive mechanism in which market participants are encouraged to play a tit-for-tat-like strategy. This simple mechanism is optimal even when the set of market partici- pants changes dynamically over time, or when some fraction of the participants may be irration
6、al. In addition to the basic model, we demonstrate how this framework can be applied to situations in which there are heterogeneous users by giv-ing several illustrating examples. This work is a first step to- wards a complete theory of incentive design for crowdsourc- ing systems. We hope to build
7、upon this framework and ex- plore more interesting and practical aspects of real online la- bor markets in our future work.IntroductionOnlinelabormarketshave emerged asapopularplatformfor matching prospective workers around the world with pay- ing work. While some online labor markets, like oDesk an
8、d Elance, focus on matching skilled laborers with relatively long-term projects, others, like Clickworker and Amazon Mechanical Turk, are designed to match workers with short, simple micro-tasks. A typical micro-task might involve cap- tioning a picture or transcribing an audio message. In principle
9、, these online labor markets for micro-tasks, or crowdsourcing markets, could revolutionize the way in which projects are completed by giving individuals immedi-ate access to large, diverse, and flexible pools of workers anyCopyright c ? 2012, Association for the Advancement of Artificial Intelligen
10、ce (www.aaai.org). All rights reserved.time of day or night. In practice, they are plagued by work- ers who attempt to exert as little effort as possible (Ipeiro- tis, Provost, and Wang 2010), and requesters who advertise spammytasksordenyworkerspaymentfortaskscompleted. Toaddressthisproblem, weadvo
11、cateincorporatinganex- plicit incentive mechanism into the design of crowdsourcing markets. In particular, we propose a class of incentive mech- anisms based on social norms (Kandori 1992).A social norm consists of a set of rules that participants are expected to follow, and a mechanism for updating
12、 participants public reputations based on whether or not they do. For example, in a crowdsourcing market, workers may be encouraged to ex- ert high effort only for tasks posted by requesters who have made payments on time in the past. Our goal is to develop a formal framework to help platform design
13、ers identify opti- mal social norms for their applications, taking into account parameters about the environment and the participants. In this work, we take into account several innate features of crowdsourcing markets:The two-sided nature. Unlike P2P systems in which all participants play similar r
14、oles, in crowdsourcing markets, the set of workers accepting tasks is typically mostly dis- joint from the set of requesters posting them.The difficulty of quality assessments.For many types of micro-tasks (for example, translation from an obscurelanguage), the quality of submitted work is difficult
15、 to determine. In some cases, requesters may need to decide whether or not to pay a worker before they are able to accurately assess the quality of his work. Even with ad- ditional time, it may not be possible to assess the quality of submitted work with certainty.Anonymity. Market participants coul
16、d potentially create new identities to erase their history and start fresh. For example, in Mechanical Turk, workers are able to create new identities with only an email address.Dynamic changes in the population. In any online com- munity, the set of participants may change over time as new individuals discover the community and current par- ticipants lose interest.The existence of irrational participants.While it is useful analytically to make the natural assumption that users are se
