Research On The Application Of Attribute Reasoning In User Profile System

Fast-growing Internet technologies have changed people's production methods and lifestyle,and emerging technologies such as big data,cloud computing,artificial intelligence,and block chain influence people's life.The social network service formed by "Internet?sociality" ignores the physical distance between people and changes the way people communicating with and making friends.Users share moods,post comments,and search for information on social networking platforms,which generate user behavior data.Software developers can use these data to make user portraits,understand user characteristics,and optimize software to provide better service for users.User portrait are widely used to optimize production design,accurate advertising,design smart recommendation services,etc.The more accurate the user portrait,the more significant the role can be.However,some criminals can also get the user data,and might do some malicious things,after analyzing the user's behavior.Therefore,user hides his characteristics in order to protect his privacy.But it makes it difficult for software developers to design and develop personalized services.In this paper,a prediction method based on attribute graph is proposed.It uses the information of training data to infer the unknown attributes of users.And a user portrait system based on Jianshu's user data is designed.The attribute graph inference method is applied to the user portrait system to improve the accuracy of the user portrait.The main work of this paper is as follows:(1)It analyzes the application of user portrait,and the problems faced by constructing user portrait.And the existing research work about user portrait is summarized.At the same time,it analyzes the privacy problems caused by user data leakage in social network platform.Finally,it summarizes the research methods of attribute privacy inference.(2)An attribute inference method based on attribute graph is proposed,which is described from several aspects such as algorithm model and numerical experiment verification.(3)In this paper,it designs a user portrait system based on Jianshu's user data.The user portrait system includes a single user portrait function and a batch of user portrait function.And the single user portrait includes a direct portrait function and a inference portrait function.(4)Applying the attribute graph-based attribute inference method to the user portrait system,the accuracy of the user portrait system is improved.