The remarkable capacity to derive insights from a graph is invaluable for risk assessment, primarily because when a risk materializes, it unfolds through a series of interconnected events that culminate at a specific point in time. Philosophically, an individual entity alone may not perceive the risk until it becomes part of a larger network of entities, each contributing to the comprehensive risk assessment
process.
The difficulty in creating a graph lies in two aspects: firstly, identifying connections and linking them to entities, all while ensuring fast performance during both the model-building and prediction phases.
The Seenity system adeptly handles this by swiftly generating a graph for each record, both during model creation and in the risk assessment process. However, the true complexity arises from deriving meaningful risk insights from the established connections within the graph. When calculating the significance of a relationship, there are several factors to consider: firstly, the essence of the relationship and the degree of correlation between the entities being assessed for risk. Secondly, it’s essential to evaluate the proximity of the relationship and utilize the distance between entities to gauge the level of risk. Additionally, it’s crucial to examine the quantities involved, including multiple connections of the same type. If a particular entity is closely associated with numerous instances of the same entities, it indicates a strong similarity.
In practice, this is precisely what the Seenity system excels at. Among other capabilities, it dynamically generates a graph, extracting meaning and calculating risk at two critical junctures: during model construction and risk assessment.
