Predicting Online News Authorship by an Authorship Embeddings Space Method

In this paper, we study the problem of authorship identification in online news data. Most of the existing approaches predict authorship via feature engineering, which cannot focus on important attributes. We designed an authorship identification method named Authorship Embeddings Space model (AES) to predict the online news authorship between online news and authors. First, we propose an authorship space to represent the deep semantic relationship of news content. Second, we use an embedding learning method to perform the relationship between authors and news. Finally, we formulated an authorship prediction algorithm to identify the news authors based on the authorship embeddings. Experimental results on the online news dataset reveal that the AES model outperforms the baseline models.