The Research Of Personalized Task Searching Method For Crowdsourcing

In recent years,crowdsourcing has gained wide attention in various fields.A new working mode arises along with the rapid development of crowdsourcing platform.As the number of tasks on crowdsourcing platform increases rapidly,people can find more kinds of tasts which need skillful workers to deal with on crowdsourcing platform.But task matching becomes a rough spot,because works are different in professional background,skills and interests.On the current crowdsourcing platform,works need to search and select the task by themselves.So,in order to guarantee accuracy of task searching,it is of great significance to make proper use of personalized search technology.The function of personalized search technology is to estimate the score of the items in intial search results by anaylse history behavior of users,and then reorder the list according to its score,and return a new result list to users.A personalized task search method based on matching degree is proposed by comprehensive analysis of the characteristics of crowdsourcing working mode and the current researches on crowdsourcing task searching and personalized technology as well as existing problems.This method builds activity logs according to history data of workers,obtains workers interests according to task query log,mines skills according to task interaction log,analyse how to estimate interest rate according to reading time,times of mouse click and scrollbar dragging and how to estimate skill rate according to the history task performance,then calculates the matching rate.Vector space model is improved by weighting task feature items according to items' location and blending the task types and matching degree to model to describe workers' preference.To guarantee the real-time performance of workers' model,model updating method is proposed.Finally,according to the worker preference model,a score evaluation function for new task text is provided to reorder the initial search results.In order to verify the validity of the research method,the personalized search method is applied to the ZHUBAJIE crowdsourcing platform.The personalized search method is verified by collecting the historical data of some workers.The experimental results show that this method outperforms historical ZHUBAJIE crowdsourcing platform search result.