Affiliation Network Model’s Constructing And Characteristics Analysis Based On Kaggle Crowdsourcing Platform

With the development of information technology, a new business model, crowdsourcing, has developed rapidly. Crowdsourcing, as a kind of social network, is currently widely used as an online, distributed problem-solving mode. Crowdsourcing, an open and sharing platform for knowledge cohesion and mutual cooperation, should mine more valuable information, which inspires users to generate knowledge and promotes all participants. So which factors influence user of the platform to participate the contest, and which factors can attract participants to continue following up. It can also give the participants the chance to share information and improve the decision.This paper starts from Kaggle.com, a typical crowdsourcing platform abroad, tracks and grabs user’s participation records in the platform, which include team ID, member ID, submission time and score,30142 records in total. Here we emphasize on the team ID and member ID, Then we use affiliation network model to build participants relationship network in crowdsourcing platform. From constructed model view, this paper determines evaluation indicators for measuring effects’importance degrees, combines metric method in complex network, and analyses that in crowdsourcing network, getting the following results. Users’ participation behavior follows long tail law, that most people’s participation frequency lower, and less people will continue follow up race; while, in formed cooperation network, less participants associate with other participants directly or indirectly, and most participants participate race alone or some participants cooperate with others; In participants directly or indirectly formed network, it also exists with the phenomenon of small world effect, that within independent groups, it also exists more tiny groups, that contact through several key participants. Grasping these key players’characteristics will be able to improve the participants’enthusiasm effectively. Furthermore, there are a huge of small connected components, which theyThrough this research, we can not only understand the bipartite networks, but also conclude the special characteristics about cooperation networks on crowdsourcing platform, then providing some advice on running crowdsourcing model. It also helps to guide the participants to improve their performance effectively, making the crowdsourcing show its nature for letting users to create labor force and collecting crowd intelligence.