Hybrid-Computing Elements: A Multi-sourcing Model for Managing Crowdsourcing Software

Abstract

The expression "hybrid-computing elements" denotes a category of computing elements where human-based and machine-based computing elements complement each other. The aim from such hybridity is to support human tasks. For example, TopCoder could be used to develop hybrid computing elements to be used in crowdsourcing software. Developing human-based computing elements is a more complex process than developing machine-based computing elements, as the task must be "structured" well as the Turing machines. This leads to the most difficult question of the "unknown unknown" requirements and having to deal with more macro-sociological factors on tasks than formal languages usually work with. In this paper, we model the hybridity using two types of computing elements: the humanbased, with whatever purposes the crowd envision, and the machine-based, which is used to develop it. We present a new framework to help provide such hybridity. It identifies the underlying building-blocks by using a biological metaphor. We call these building-blocks the "genes" of collective intelligence systems, the conditions under which each gene is useful, and the possibilities for combining and re-combining these genes to harness crowds effectively. Employing an analogy from biology where the operator relating the crowd and the parameter to be social phenomena is assumed to belong to semi-algebraic sets. We evaluated our framework by developing a business application through crowd work. The primary result was completed well for managing the crowdsourced software.