Research On Task Recommendation Algorithm For Crowdsourcing Design Platform

With the improving of the level of information technology,the transformation of the production to service-oriented manufacture is the inevitable trend in our country.The crowdsourcing design pattern provides a new solution to the fast convergence and effective application of manufacturing resources as a business mode of collecting the global resources.In the existing crowdsourcing platform,the users have to go through a lot or do artificial search to find suitable task,which has made information overload,inefficient operation of platform,etc.In order to solve these problems,this paper has proposed content recommendation algorithm based on task-user model and based on social network for competitive crowdsourcing and collaborative crowdsourcing,and realized the matching process of task and user,it has improved the accuracy of task recommendation algorithm and reduced the problem of information overload,etc.(1)For the features of competitive crowdsourcing design platform,content recommendation algorithm based on task-user model is proposed.This paper has expounded the building methods of task and user model,the matching algorithm,and the solution of the problem of user cold start.We consider the influence factors from four dimensions for more accurate recommendation.The experimental result shows that this algorithm could generate personal recommendation list effectively,and solves the problem of effective utilization of user resources,and validates the relationship between model element weight and the results of task recommendation.(2)In view of the characteristics of collaborative crowdsourcing design platform,the content recommendation algorithm based on social network analysis is proposed.The framework of the algorithm and the functions of its core components is proposed.It considers the influence the relationship between users,which will influence the algorithm,and it uses the theory of social network analysis to search neighbor users.The experimental result shows that the high accuracy of this algorithm.(3)In order to solve the dynamic problem of crowdsourcing platform,we proposed a task recommendation system based on Multi-Agent.This paper builds the model of Multi-Agent task recommendation system,and expounds the function of each agent,the interaction relations between agents.The related algorithms of users' ability level are proposed.Using the simulation software Net Logo to verify,the result shows that the recommendation system can play a promoting effect to the promotion of users' ability level,and prove task recommendation based on Multi-Agent can solve the dynamic problem of crowdsourcing platform.