Depression Tendency Detection of Weibo Users Based on Knowledge Graph

The detection of depression tendency of Weibo users has become a research hotspot today. This paper proposes a depression tendency detection method based on Knowledge Graph. The ultimate goal is to identify Weibo users with depression tendencies more accurately. This method filters the text on Weibo User Depression Detection Dataset(WU3D) and obtains the lexicon ontology of emotional words through manual annotation. After merging with the Chinese emotional lexicon ontology of Dalian University of Technology, the same data is removed. The Emotional Knowledge Graph is constructed by emotional words and the relation between them. The text features are extracted by Emotional Knowledge Graph and the pre-trained BERT+BiLSTM model separately. The extracted features are fused by concatenating. The results are obtained by Softmax. The experimental results show that the proposed method has better performance in Accuracy, Precision, Recall and F1 score.