Research on Gender Prediction for Social Media User Profiling by machine learning method

Abstract

This paper aims to propose a gender prediction method that integrates social media users' sentiment. This method is used to predict the gender attributes of users, so as to realize the research on the construction method of social media user portraits. Previous studies on gender prediction have done little analysis of sentiments. This paper has used the idea of transfer learning to analyze users' sentiments and integrated sentimental features into the existing machine learning, and hence, shows superior performance in terms of accuracy as compared to other methods. This paper mainly uses machine learning method to realize the construction of social media user profiling. The gender attribute is studied. Firstly, feature extraction is carried out for text data of media users. Then the idea of transfer learning is used to analyze the user's sentiment and integrate the sentiment characteristics into the existing machine learning. Finally, five prediction methods, i.e. Logistic Regression(LR), Naïve Bayes(NB), k-Nearest Neighbor(KNN), Random Forest(RF) and Support Vector Machines(SVM) is used to predict the gender of the fused sentiments. The results show that The gender prediction effect after sentiment fusion is better than that before. and the accuracy is increased by about 2.1% on average.