Automatic Recognition And Analysis Of Weibo User’s Age Range

Weibo as one of the most popular socialnetwork sites, has become an important platform for people to share, exchange, obtain and spreadinformation in their lifes. There are about 100 millions people to login in Weibo every day. We can analyze the characters of several groups of age ranges in more detail by knowing the user ’s age, for more in-depth mining the content of tweets, thus, to obtain the huge commercial value of large amounts of data which is generated by users of Weibo. However, most of users do not have the label of age in their profiles, not only that, with the close of the API in Weibo, the data acquisition is more and more difficult, which lead to a decline in the practicality of the model of automatically judge the users ages range by their profile, tweets and so on.The purpose of this paper is to construct a model that can automaically identify the age range of a Weibo user, just relying on his tweets. Then we apply the model to discern some users age ranges to analyze the characters of kinds of age range users on Weibo. In this paper, we chose 5466 users by a artifical method as the training samples of the model and 950 000 users in Weibo as the late analysis samples. Then, we applied four machine learning algorithms to construct some models, and compared to the results of models. The research content of this paper mainly includes the following:Firstly, relying on the text of tweets of Weibo users, we constructed a model for recogniting antomatically the age range of a user. We selected randomly 5466 Weibo users by artificial method who had date of birth in the profile, and obtained their tweets. With words, emoticons and punctuation marks as inputs, four classifiers were trained to classify users into four age groups. And then the best classifier, logistic regression model, was chosen as the model in this paper.Secondly, the paper analyzed the characters of different age groups of Weibo users. In the paper, there are 0.95 millions users been selected in Weibo, and their ages were estimated with the age automatic detection model. We compared the four groups in the aspect of population, activity time, number of friends and topics in tweets, to find that they were different in many ways.This paper proposes an automatic model to detect the age range of Weibo users based on tweet texts only, which means that the model can be applied to age detection in other platforms. The model itself and our analys is on Weibo user in different age groups are helpful in policy, economy, law s and so on, and provide information like age for other researches based on Weibo.