Towards automated financial market knowledge graph construction

The prevalence of financial news on the internet has made it easier for investors to access information. However, the fast-changing nature of the financial market and the time-consuming task of sifting through articles can be overwhelming. To address this issue, a framework has been developed and is proposed to automatically construct and update a knowledge graph (KG) for financial market information. The KG stores relational information between entities in a directed graph format, providing a graphical visualization that allows investors to examine complex relationships between entities that play a role in the stock market. The framework involves five main phases: scrapping online articles, triples extraction, coreference resolution, predicate linking, and entity linking. The precision rate achieved by the framework is 27.69%, with a recall rate of 7.14% and an F-1 score of 0.1136 in extracting correct information from articles and integrating it properly into the KG.