Personal Ranking of Projects for Workers in Crowdsourcing Systems

In many huge services, where there are many users and items, sooner or later the task of recommendation/ranking arises. In particular, there is a crowdsourcing system. However, unlike classic learning-to-rank problems such as movies recommending, crowdsourcing ranking has the peculiarity that we must take into account the interests of not only users, but also the projects that users perform. Thus, within the framework of crowdsourcing, the learning-to-rank task takes a new form of fair worker distribution on projects with the preservation of freedom of choice. We cannot explicitly assign each worker to the desired project, but we can build a ranking model that will gently recommend projects to workers, taking into account both their personal skills and the interests of the projects on the quality of the performers' work. In this paper, we investigate the problem of learning-to-rank in the crowdsourcing in the context of the problem of a two-way market with the possibility of "many-to-many" communication and try to build a qualitative ranking model that takes this problem into account.