Designing Incentive for Cooperative Problem Solving in Crowdsourcing
Huan Jiang · Kyoto University Research Information Repository (Kyoto University) · 2015
The goal of this thesis is to design incentives for workers in crowdsourcingbased cooperative task solving, with the aim of improving the quality of task solution. Crowdsourcing is an online, distributed task-solving and web-based business model that has emerged in recent years. It is now admired as one of the most lucrative paradigm of leveraging collective intelligence to carry out a wide variety of tasks with high complexity. This very success is dependent on the potential for replacing a limited number of skilled experts with a multitude of unskilled crowds, by decomposing the complex tasks into smaller pieces of “micro-tasks” (or subtasks), such that each subtask becomes low in complexity, requires little specialized skill, time and cognitive effort to be completed by an individual, and models a step or operation needed in the sequence of producing a final solution to the complex task. However, it is also possible that crowdsourcing-based work will fail to achieve its potential. When facing a geographically distributed workforce who has various knowledge domains and levels of abilities, task requesters always have difficulty to ascertain the quality of the submitted result. Moreover, the strategic behaviors of the rational workers who aim at maximizing their own utilities could have great impact on the quality of the task solution. Drawing on incentive design perspective, this thesis addresses following issues in crowdsourcing with respect to task-solving model construction, worker’s behavior analysis, incentive design and experimental implementation. 1. Designing efficient task decomposition strategy for solving complex