Crowdsourcing Task Assignment Method Based On The Attribute Space Of Pairwise Associations And Attribute Selection

Participant and Crowdsourcing tasks are two elements in crowdsourcing task assignment. Each participant can provide the knowledge and skills with different attributes characteristics, and each task requires knowledge and skill needs with different attributes characteristics. In the network environment, How appropriate crowdsourcing tasks are assigned to the appropriate participants both have the opportunities and challenges, the opportunities include revealing their ability characteristics and task demand characteristics by UGC, the challenges include massive and high-dimensional attributes data and their interactions.Different with the traditional “one to one “assignment degrees between the participants and task, there are many-to-many task assignment among participants and tasks in crowdsourcing task assignment. In traditional task allocation model, some ignored the attributes characteristics of participants and tasks, and some assumed that these attributes characteristics are independent of each other. in fact, crowdsourcing task assignment should consider the certain interrelated relationship among any demand characteristics between pairs of tasks and, as well as the interrelated relationship among the ability characteristics among the participants.In this paper, we propose a new crowdsourcing task assignment method that takes into account the attributes of tasks and participants as well as associations between the attributes.The main work of this paper include:(1) the attribute space is constructed based on the association between the paired attributes;(2)In this space, the similarity degree between any pair of crowdsourcing tasks and any pair of participants are calculated. According to the similarity degree and the bipartite graph of crowdsourcing task VS participants, optimistic crowdsourcing task assignment model is constructed. According to result of crowdsourcing task assignment degree, determined the main attribute characteristics of influencing crowdsourcing task assignment utility and constructed the attribute selection model.Simulation results show that the assignment is affected by the scale of tasks and participants. As adaptive dimensions of attributes according to related weight thresholds are included, the assignment results are improved. Case analysis indicates that the model possesses a certain assignment validity compared with the optimal assignment in traditional Euclidean space. the optimal task assignment model in PAAS can reveal the assignment degree between participants from certain communities and the crowdsourcing tasks related to certain domains.The task assignment model and attribute selection model based on ability characteristics of participants and demand characteristics of crowdsourcing tasks in the network environment not only enrich the theory of crowdsourcing task assignment and attribute reduction of large data network but also contribute to assignment the task to the participant in practice and improve the unity of crowdsourcing task assignment in the large-scale network.