Research On Crowdsourcing Talent Selection Based On Supervised Learning

With the great development of Internet technology,crowdsourcing as a novel task trading mode under the Internet environment,has been widely used by various demand holders or merchants to solve their task needs.As a result,a batch of crowdsourcing platform sites with crowdsourcing pattern have been developed and attract a large number of task publishers and task-seekers,who are the crowdsourcing service providers and receive the amount of compensation by completing tasks on the crowdsourcing platform.In the face of a large number of virtual crowdsourcing service provider information,how to select the service provider with the high quality of the release task from these crowdsourcing service providers quickly and accurately has become an urgent problem to be solved.Besides,the complexity of task type on the crowdsourcing platform also makes it more difficult to solve the problem of selecting service provider.At present,the research about crowdsourcing service provider selection and talent screening on crowdsourcing platform is mainly based on empirical research to explore the influencing factors that affect the participation of crowdsourcing service providers or to explore the influencing factors of crowdsourcing service provider selection based on the management theory of talent screening.Few scholars have set up a model to complete the selection of influencing factors for crowdsourcing service providers to explore the whole process of landing service provider selection example analysis.This paper solves the selection of crowdsourcing talent as a two-stage classification problem,we construct a packet service provider classification selection model based on supervised learning algorithm,and in this paper,based on three classification methods,we train model supervised and apply the trained optimal model to the selection of crowdsourcing talent in ZhuBajie website.The current classification idea is widely used in the online community,for example,some scholars use the numerical features on the platform to evaluate the answer quality of the Q&A website,or to classify users on the online medical platform based on online comment data,but few people discuss the application of classification algorithm in crowdsourcing website,Crowdsourcing website has rich online comment data,which provide a lot of research space for the application of classification algorithm.The research ideas of this paper include some:firstly,based on the previous research and theory to explore the influencing factors of crowdsourcing service provider selection,combining the winning competence theory with the crowdsourcing characteristics based on the crowdsourcing platform,Contracting parties mentioned(crowdsourcing service provider)three main body,build crowdsourcing service provider selection of influencing factors index system,and extract feature data based on the index system;Based on Decision tree,KNN and Bayesian classification algorithm,they have supervised training classifier,using the trained classifier to screen the crowdsourcing service provider,and then determine whether the better category in the classification results is the actual winning service provider,the inferior category is the actual non-winning service provider Finally,based on the determination results,the accuracy of the packet service provider classification selection model based on supervised learning algorithm is calculated,and the influence of different feature factor combinations on the classification selection results of crowdsourcing service providers is discussed.This paper takes ZhuBajie website as an example to analyze,the experimental results show that the selection results of crowdsourcing service providers are excellent based on the structure combines contracting parties(crowdsourcing service providers),crowdsourcing platform and task design.Secondly,the classification selection model of crowdsourcing service providers based on supervised learning algorithm in this paper has better results in accuracy and recall rate.Among them,the model of Bayes algorithm has the best application effect,and the best service provider of model selection basically conforms to the winning service provider selected by the contractor in the actual data,the model effectively improves the accuracy of the selection of crowdsourcing service providers and verifies the effectiveness of the model in practical application.