Risk identification and management through knowledge Association: A financial event evolution knowledge graph approach

The spread and evolution of financial risk events have endangered the operation of enterprises and financial institutions, and even strike the whole social credit system. To clarify the path of risk events and identify key risk sources by event association, we propose and construct a novel event-driven knowledge graph called “Financial Event Evolution Knowledge Graph (FEEKG)”, with a multi-layer structure of “entity-event-risk”. In the entity layer, the subgraph of about 112,000 entities and 78,500 relationships (including 6 types of entities and 12 types of relations) is constructed based on the event extraction algorithm and topic model. In the event layer, the evolution relationship scores higher than 0.2 is selected as the edge of the event evolution subgraph. The risk layer takes 23 types of topic risk events and 12 types of second-level risk types as nodes and risk transition relations as edges, to generate the dynamic evolution probability subgraphs of topic risk events and risk types. Driven by risk events, we also analyze the characters and rules of entity correlation, event evolution, and risk transmission based on FEEKG; propose a new method for timely mining potential related risk entities and effectively summarizing the regularity of risk evolution and transmission and provide a new perspective for enterprises and financial institutions to find the root of risks and formulate an effective risk management decision in time. © 2024 Published by Elsevier Ltd.