DataTerminology.comNEW
Also known as: Default Reasoning, Defeasible Reasoning
Logical reasoning that permits withdrawing conclusions when new information contradicts previously held defaults or assumptions.
DESCRIPTION
Non-monotonic reasoning models defeasible, exception-rich knowledge common in real-world domains. New evidence can invalidate prior conclusions, reflecting human-like default reasoning. Formal systems include default logic, circumscription, and autoepistemic logic. Applications span planning, diagnosis, and knowledge representation with uncertainty. Implementations often trade completeness for practicality, requiring careful governance to avoid unstable behavior.
ALTERNATIVE DEFINITIONS
Stanford Encyclopedia of Philosophy
Nonmonotonic logics are formalisms in which the addition of new premises can invalidate previous conclusions.
Captures the defining property of nonmonotonic inference.
AI Magazine (AAAI)
Nonmonotonic reasoning addresses the need to draw conclusions that may have to be retracted when additional information is obtained.
Frames nonmonotonicity as practical need in AI systems.
RELATED TERMS
Default Logic Circumscription Autoepistemic Logic Reasoner
COMPETING TERMS
RELATIONS
- Broader: Reasoning
KNOWLEDGE AREAS
AI Mathematics Semantic & Knowledge Modeling
TARGET AUDIENCE
EXTERNAL SOURCES
- Wikipedia - Non-monotonic logic
- Stanford Encyclopedia of Philosophy - Comprehensive overview of nonmonotonic logic
- AAAI - AI Magazine article on nonmonotonic reasoning
TOP QUESTIONS
Why use Non-monotonic Reasoning in AI?
Real-world knowledge includes exceptions and defaults; non-monotonic formalisms model revisable conclusions more naturally.
Category: Relevancy
What are key Non-monotonic Reasoning formalisms?
Default logic, circumscription, and autoepistemic logic are classic approaches.
Category: Overview
What risks arise with Non-monotonic Reasoning?
Complex semantics, potential instability, and harder debugging compared to monotonic systems.
Category: Risk
How does Non-monotonic Reasoning differ from probabilistic inference?
It retracts conclusions via logical rules, not probabilistic updates; they can be combined pragmatically.
Category: Misunderstanding
How is Non-monotonic Reasoning implemented?
Specialized reasoners, rule engines with defaults, or hybrid systems embed defeasible rules and priorities.
Category: Requirements
What is a hot topic around Non-monotonic Reasoning?
Combining symbolic defaults with statistical models for robust decision-making.
Category: Topic
How do you validate Non-monotonic Reasoning behaviors?
Define scenarios, test exceptions, ensure consistent priority handling, and monitor changes under new facts.
Category: Management
Where to learn Non-monotonic Reasoning?
Start with Stanford Encyclopedia entries and foundational AI Magazine articles.
Category: Learning
What’s a practical use case for Non-monotonic Reasoning?
Policy engines with exceptions and context-dependent rules in compliance or access control.
Category: Example