BayesL: Towards a Logical Framework for Bayesian Networks
This addresses the need for more efficient and automated reasoning tools in probabilistic modeling, though it appears incremental as it builds on existing Bayesian network methods.
The paper tackles the problem of specifying, querying, and verifying Bayesian networks by introducing BayesL, a logical framework that enables versatile reasoning and what-if scenario evaluations without manual model modifications.
We introduce BayesL, a novel logical framework for specifying, querying, and verifying the behaviour of Bayesian networks (BNs). BayesL (pronounced "Basil") is a structured language that allows for the creation of queries over BNs. It facilitates versatile reasoning concerning causal and evidence-based relationships, and permits comprehensive what-if scenario evaluations without the need for manual modifications to the model.