Promoting Financial Market Development-Financial Stock Classification Using Graph Convolutional Neural Networks

According to surveys, one in seven people in China are involved in stock trading, and the role of stocks in the global economy is growing. In terms of the entire stock market, the Chinese market alone has over 4000 stocks. Faced with a chaotic and diverse assortment of stocks, it is necessary to categorize them. On the one hand, it can facilitate stock market research, and on the other, it can make it easier for stockholders to purchase shares. The graph convolutional neural network-based SK-GCN model developed in this paper delivers excellent results in the categorization of stock classes. This model employs two layers of convolutional layers and activation functions to effectively categorize stocks by incorporating external nodes to expand stock features and drawing inspiration from short text classification. This strategy is highly innovative and produces promising outcomes. In this paper, we constructed the dataset by crawling the information of all stocks listed on the GEM of Oriental Fortune website. We achieved an accuracy of 83.04% and a macro-F1 value of 0.8303 under the assumption of small sample training, and its classification effect is significantly superior to other classification models.