Construction Research and Applications of Industry Chain Knowledge Graphs

Research on listed companies is an important part of stock analysis. This study proposes an automatic construction method of the knowledge graph for the financial industry chain, which can better track and study the relationship between the production and operation of listed companies from the perspective of the graph. To solve the problems of high expert labor costs, late update and maintenance, and unstandardized and label-lacking datasets during the construction of vertical domain knowledge graphs, this study conducts knowledge extraction of unstructured text by integrating two dependency parsing methods, phrase structure trees, and dependency parse trees. Furthermore, this study uses automatically labeled datasets to train a deep learning-based named entity recognition model. This method, which integrates the two abovementioned methods, improves normalization ability and allows for the automatic construction of a financial industry chain knowledge graph.