Dynamic Max Algorithms in Crowdsourcing Environments
Dynamic Max Algorithms in Crowdsourcing Environments
复制标题
众包环境中的动态最大算法
DOI：
10.1145/2339530.2339707
复制
发表时间：
2012
期刊：
影响因子：
--
通讯作者：
H. Garcia
中科院分区：
文献类型：
--
作者：
Petros Venetis;H. Garcia
我们的工作研究了在众包环境中从集合中检索最大项目的问题。我们专注于锦标赛算法，例如可以选择与给定人员匹配的最佳Facebook个人资料或描述给定餐厅的最佳照片。锦标赛算法可以通过参数进行调整，例如前2名投票结果之间的期望差异，以及要求执行特定任务的最大人数。我们提出了一个策略，选择适当的比赛参数，试图保持货币成本和延迟在最低限度，同时有质量保证。对于我们的实验，使用来自心理测量学的人类模型（Thurstonian模型），我们提供了我们的策略在选择适当的比赛参数的有效性的见解，并将我们的技术与以前的比赛调整工作进行比较。
Our work investigates the problem of retrieving the maximum item from a set in crowdsourcing environments. We focus on tournament algorithms that can for instance select the best Facebook profile that matches a given person or the best photo that describes a given restaurant. Tournament algorithms can be tuned with parameters such as the desired difference of votes between the top-2 voted outcomes, and the maximum number of humans asked to perform a particular task. We propose a strategy for selecting appropriate tournament parameters that attempts to keep monetary cost and latency at a minimum while having quality guarantees. For our experiments, using a human model derived from psychometrics (the Thurstonian model), we provide insights on the effectiveness of our strategy in selecting appropriate tournament parameters and compare our techniques with previous work on tournament tuning.