Formalizing complex normative decisions with predicate logic and graph databases

Abstract

This paper argues that the critical work in deontic reasoning is better done in the knowledge representation rather than the reasoning of a normative system. It describes a way to formalize complex normative decisions using predicate logic and graph databases. Simple norms can be mechanized with IF/THEN statements. While often expressed in deontic logic, such statements can be expressed in simpler predicate logic. More complex normative decisions require the ability to make decisions where there are multiple clashing duties. Such decisions could be formalized in graph databases that express state-act transition relations, causal relations, classification relations and evaluation relations. When formalizing complex normative decisions it is more powerful and practical to draw upon concepts from multiple moral theories rather than restricting the system to a single theory. A normative system with extensive knowledge representation of complex relations might be able to pass a series of reasonable person tests. Passing such tests rather than implementing a particular moral theory should be the main design aim of normative systems.