A Trusted Task Fitness Based Crowdsourcing Team Formation Mechanism In Social Network

With the development of the Internet,crowdsourcing has become familiar and has been used by more enterprises and groups.The crowdsourcing tasks can be divided into simple tasks and complex tasks.Simple tasks are inseparable and complex tasks can be divided into multiple sub-tasks according to skills,hiring multiple people to cooperate.Therefore,how to form a professional team becomes a problem to solve for crowdsourcing teamformation.In the current crowdsourcing team formation researches,people have proposed a lot of different schemes to stimulate workers' true quotation or improve group efficiency.However,people ignore the rationality of workers who want to maximize personal utility,the difficulty of tasks and workers' fitness for the tasks.What's more,agents in the network are not always willing to cooperate and there are even some malicious workers who participate teams maliciously in order to collapse the system.Therefore,this thesis first introduces the concepts of task difficulty and self-estimated fitness,and proposes a self-estimated fitness and negotiation based crowdsourcing team formation mechanism(FNTF)in social networks.However,there are malicious attacks existing in the network,introducing task fitness and using the dynamic changes of the network in the process of negotiation,a trusted task fitness and negotiation based crowdsourcing team formation mechanism(TFNTF)is proposed.With the help of task fitness update,negotiation based trusted worker selection,rewards and punishments,etc.,TFNTF effectively estimates freelancers' task fitness,isolates malicious workers,promotes team formation and the establishment of cooperative relationship among normal workers.Through the experiments,this thesis proves that FNTF and TFNTF can improve the social utilty and the success rate of tasks' execution when tasks require workers' feasible fitness.Meanwhile,when malicious attacks exist,TFNTF can effectively resist attacks of five attack models,improving social utility,ensuring similar or higher team formation efficiency and higher task execution success rate,reducing social losses due to tasks' failures.