A Bayesian network is a probabilistic graphical model that maps relationships between variables using nodes and directed edges. Each node represents a variable and the edges between them show how those variables influence one another. Together, they form a structured way to represent uncertainty and work through it logically.
The model calculates the likelihood of unknown variables based on what is already known. It can also update those probabilities as new information comes in, making it useful in situations where data is incomplete or constantly changing. This reasoning process is closely related to how bias in machine learning is managed, since the model's starting assumptions directly shape its outputs.
A core strength of Bayesian networks is interpretability. Their graphical structure makes tracing how a conclusion was reached straightforward, which is useful when transparency matters.