A Social Bots Detection Model Based on Deep Learning Algorithm
Heng Yu Ping, Su‐Juan Qin · 2018
With the development of the Internet, social bots are increasingly spreading on social platforms. Therefore, an effective detection algorithm is demanded to detect these social bot accounts that endanger social networks. In this paper, a social bots detection model based on deep learning algorithm (DeBD) is proposed. The model mainly includes three layers. The first layer is the joint content feature extraction layer, which focuses on the feature extraction of the tweets content and the relationship between them. The second layer is the tweet metadata temporal feature extraction layer, which regards the tweet metadata as temporal information and uses this temporal information as the input of the LSTM to extract the user social activity temporal feature. The third layer is the feature fusing layer, which fuses the extracted joint content features with the temporal features to detect social bots. To evaluate the effectiveness of the DeBD model, we conducted experiments on three different types of new social bot data sets from the real world and the experiment results also demonstrate the effectiveness of our proposed model.