众包任务分配算法的改进与应用

发布时间：2018-05-03 03:19

本文选题：众包市场 + 任务分配　；参考：《大连海事大学》2017年硕士论文

【摘要】：众包市场作为一种利用大量互联网工人来廉价收集数据的方法在研究者之间变得非常流行。然而,众包对于工人管理较弱,如何将多样性任务合理分配给拥有不同未知技能工人仍然是一个亟待解决的问题。多重任务分配算法(Dual Task Assigner,DTA)是一种解决众包环境下多样化任务分配给未知技能众包工人问题的新型任务分配算法。不过DTA算法对不诚实的众包工人识别较差和对高技能等级的众包工人利用不够充分。为此本文将改进DTA算法使其更为有效的分配任务。本文首先提出工人选择算法来选出诚实的众包工人进入到后续的任务分配中,并与众包平台的自动检索系统(Automatic Retrieval System,ARS)进行对比发现工人选择算法能够有效的选出诚实的众包工人,降低不诚实的众包工人对后续任务分配的影响。其次,基于工人选择算法的迭代结果计算众包工人的技能等级值,并使用人工蜂群算法(Artificial Bee Colony,ABC)计算每个任务的基线权重值。然后,针对DTA算法分配策略单一的问题,本文新增高技能等级的众包工人多分配任务和技能等级不足的众包工人不给予任何任务分配两种分配策略,这样可充分利用高技能等级的众包工人,并确保最终任务结果拥有较高的质量。最后,基于xTurk收集的真实实验数据对改进后DTA算法进行实验性分析发现改进后DTA算法的表现优于DTA算法和随机分配算法。在本文的最后,我们使用Python和Django框架搭建一个简易的众包网站,并将改进后DTA算法的分配策略应用于该网站来实现合理的任务分配,通过简单的交互过程使用户加深理解改进后DTA算法的任务分配过程。
[Abstract]:Crowdsourcing markets have become popular among researchers as a way to collect data cheaply using a large number of Internet workers. However, crowdsourcing is weak for workers, and how to distribute diversity tasks to workers with different unknown skills is still an urgent problem to be solved. Dual Task assignment algorithm (DTA) is a new task assignment algorithm for solving the problem of multi-task assignment to workers with unknown skills in crowdsourcing environment. However, the DTA algorithm has poor identification of dishonest crowdsourcing workers and inadequate use of high skill level crowdsourcing workers. Therefore, this paper will improve the DTA algorithm to allocate tasks more effectively. First of all, this paper proposes a worker selection algorithm to select honest crowdsourcing workers to enter the subsequent task assignment, and compared with automatic Retrieval system ARSs, it is found that the worker selection algorithm can effectively select honest crowdsourcing workers. Reduce the impact of dishonest crowdsourcing workers on subsequent assignments. Secondly, based on the iterative results of the worker selection algorithm, the skill level of the crowdsourcing workers is calculated, and the baseline weight of each task is calculated by artificial bee colony algorithm (artificial Bee colony). Then, in order to solve the problem of single allocation strategy of DTA algorithm, this paper adds two assignment strategies: high skill level crowdsourcing workers and low skill level crowdsourcing workers do not give any assignment strategy. This makes full use of high skill level crowdsourcing workers and ensures high quality of final task results. Finally, the experimental analysis of the improved DTA algorithm based on the real experimental data collected by xTurk shows that the performance of the improved DTA algorithm is better than that of the DTA algorithm and the random assignment algorithm. At the end of this paper, we use Python and Django framework to build a simple crowdsourcing website, and apply the allocation strategy of the improved DTA algorithm to the website to achieve a reasonable task allocation. Through the simple interactive process, the user can understand the task assignment process of the improved DTA algorithm.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18

【相似文献】