[Paper Review] Toward a Characterization of Uncertainty Measure for the Dempster-Shafer Theory

David Harmanec|arXiv (Cornell University)|Feb 20, 2013

Multi-Criteria Decision Making10 references19 citations

TL;DR

This paper proposes a set of axiomatic requirements for uncertainty measures in the Dempster-Shafer Theory, demonstrating that the recently introduced measure AU is minimal among all measures satisfying these conditions. It establishes a formal characterization of uncertainty by proving AU's uniqueness under the proposed axioms, contributing to a foundational understanding of belief function uncertainty quantification.

ABSTRACT

This is a working paper summarizing results of an ongoing research project whose aim is to uniquely characterize the uncertainty measure for the Dempster-Shafer Theory. A set of intuitive axiomatic requirements is presented, some of their implications are shown, and the proof is given of the minimality of recently proposed measure AU among all measures satisfying the proposed requirements.

Motivation & Objective

• To formally characterize uncertainty measures in the Dempster-Shafer Theory using intuitive axiomatic principles.
• To identify minimal requirements that any valid uncertainty measure must satisfy.
• To prove that the measure AU is minimal among all measures fulfilling the proposed axioms.
• To contribute a foundational characterization of uncertainty in belief function theory for AI applications.
• To provide a rigorous axiomatic basis for future development of uncertainty measures in evidence theory.

Proposed method

• The authors define a set of intuitive axiomatic requirements for uncertainty measures, including consistency with belief function structure and monotonicity.
• They analyze logical implications of these axioms, showing their necessity and coherence in modeling uncertainty.
• The proof of minimality of AU involves comparing AU against all other potential measures satisfying the axioms.
• The analysis relies on formal reasoning over the frame of discernment and mass functions in the Dempster-Shafer framework.
• The method uses logical deduction and comparison to establish that no smaller measure can satisfy the axioms.
• The approach is theoretical and axiomatic, focusing on structural properties rather than empirical validation.

Experimental results

Research questions

• RQ1 What axiomatic conditions must a valid uncertainty measure satisfy in the Dempster-Shafer Theory?
• RQ2 Is the measure AU uniquely minimal among all measures satisfying the proposed axioms?
• RQ3 How do the proposed axioms constrain the form of uncertainty measures in belief function theory?
• RQ4 Can a formal characterization of uncertainty be achieved through a minimal set of intuitive principles?
• RQ5 What is the relationship between the structure of mass functions and the resulting uncertainty measure?

Key findings

• The proposed set of axioms provides a coherent and intuitive foundation for uncertainty measures in the Dempster-Shafer Theory.
• The measure AU is proven to be minimal among all measures satisfying the axioms, meaning no smaller measure can fulfill the conditions.
• The axioms imply key structural properties such as monotonicity and consistency with belief function semantics.
• The proof establishes AU as the unique minimal solution under the given constraints, supporting its theoretical primacy.
• The results provide a formal characterization of uncertainty, advancing the theoretical underpinning of evidence theory.
• The work confirms the suitability of AU as a canonical uncertainty measure within the Dempster-Shafer framework.