Research Of The Relationship Between Weibo And Television User Based On User Group

Tags are very useful to partition user into different groups.The TV industry hopes to improve the accuracy of TV user profile by tags from other source.There are abundant tags on Weibo,but as a data source to improve the accuracy of TV user profile,the relationship between Weibo users and TV users also needs to be determined.This paper studies the relationship between Weibo and TV user groups by evaluating user profile of Weibo,TV and the data based on Weibo data conversion.First,use the distributed framework Hadoop for data collection and cleaning.Then,based on Weibo and TV user data,user profiles were constructed using Word2Vec and K-Means clustering algorithms.After that,a sample domain conversion model is constructed based on the data distribution and the GAN model.This model converts Weibo data into data more consistent with the distribution of real TV users,and builds user portraits.By comparing the data obtained from the model conversion and the user profile effects constructed by Weibo and TV data,it is.found that the clustering distribution of Weibo data after model conversion is closer to the distribution of TV data.Therefore,it is believed that there may be a correlation between Weibo and TV,which supports the use of Weibo tag data to improve the accuracy of TV user portraits.