Research On User Portrait Construction Technology And Visualization Application

User portrait technology is based on the user's existing information,using machine learning or deep learning methods to mine the potential attributes of users,so that different users can be tagged with their own unique technology.Based on these tags,researchers can achieve more accurate operation for users.At the same time,thanks to the development of visualization technology,it is easy to visualize the results of user portraits,and to obtain the results of user portraits in a more intuitive way,which makes the researchers understand the data more thoroughly,and also provides convenience for the follow-up work.Based on the above information,this paper mainly completes the research of user portrait technology and the design and implementation of visualization system.Firstly,the research of user portrait construction technology based on microblog of Internet social media is carried out,and a convolutional neural network model based on deep level is established.By splicing the output of each convolution layer as the input of the next convolution layer,the model can not only make full use of the upper features in the lower convolution,but also make effective use of the interaction between context information,and also realize the flexible selection of multi-scale features.In addition,a gated calculation is added to each convolution layer.By gated calculation,the characteristic map which is beneficial to the experimental results can be selected in time.By gated calculation,the gradient descent problem arising from the calculation process can also be reduced,so that the model can flexibly increase the number of convolution layers.Secondly,the application research of user portrait construction based on precise poverty alleviation is carried out in the context of government poverty alleviation data,and a convolutional neural network method is adopted,which integrates the statistical and text features of users.Finally,a user portrait visualization system is designed and implemented using Web technology.In the design of the system,combining the characteristics of visual scene demand and data diversity,fully taking into account the visualization of multi-dimensional,multi-level and multi-type data,through Echarts and other open source databases,reasonable use of pie charts and polyline charts to display statistical data,and combined with word clouds,scatter plots,drilling technology,etc.,to achieve multi-granularity of portrait information.Visualization,personalized design of Web pages,combined with existing open source technology,to achieve responsive layout of Web pages and good human-computer interaction functions.This paper carries out comparative experiments on Sina Weibo dataset.The experimental results show that the deep-level convolution neural network model adopted in this paper has good performance.Because of the limited data available for government poverty alleviation,this paper uses convolutional neural network to carry out experiments,completes the classification operation of poverty causes of poor households.In the construction of the visualization system,this paper uses Ajax to realize data asynchronous interaction and other operations,and realizes the visualization display of user portraits.