Study on Social Media Depression Classification Method Based on Text Data

The cause of depression is not well understood by researchers at present, but its existence has seriously harmed people's health. Therefore, it is very important to quickly judge whether it is depression in today's society. In the work, the text dataset from social media is first vectorized with Bag-of-words and self-training Word2Vec, and then the text represented by Bag-of-words is trained and tested in different types of traditional machine learning classification methods. For text represented by Word2Vec, CNN and RNN algorithms are added for training. Finally, we compare and analyse the classification effects of traditional machine learning algorithms and deep learning algorithms with different text vectorization representations on text dataset, so as to find a better classification model of depression based on text dataset. The experimental results show that CNN and Logistic Regression model are better in the task of depression classification based on text dataset.