A computational approach to approximate and plausible reasoning with applications to expert systems

Area of Science:

• Artificial Intelligence
• Fuzzy Logic
• Uncertainty Quantification

Background:

• Managing uncertainty and imprecision is crucial in artificial intelligence.
• Existing methods for uncertainty modeling are diverse and often lack a unified framework.

Purpose of the Study:

• To propose a common basis for modeling uncertainty and imprecision.
• To discuss approximate and plausible reasoning schemes within this unified framework.

Main Methods:

• Unified introduction of various uncertainty measures (probability, belief functions, possibility theory, fuzzy measures).
• Modeling imprecision using possibility distributions.
• Investigation of deductive inference under uncertainty and imprecision.

Main Results:

• A framework for integrating diverse uncertainty measures is presented.
• Analysis of reasoning with fuzzy events, truth qualifications, and multi-valued logics.
• Methods for combining uncertain and imprecise information from multiple sources.

Conclusions:

• The proposed framework offers a comprehensive approach to uncertainty and imprecision modeling.
• This work provides a foundation for advanced reasoning in artificial intelligence systems.
• Addresses key challenges in managing uncertainty and imprecision in AI.