Implementing Dempster-Shafer Theory for property similarity in Conceptual Spaces modeling

Sensor Systems and Information Systems IV, American Institute of Aeronautics and Astronautics (AIAA) SCITECH Forum 2022 (2022)
  Copy   BIBTEX

Abstract

Previous work has shown that the Complex Conceptual Spaces − Single Observation Mathematical framework is a useful tool for event characterization. This mathematical framework is developed on the basis of Conceptual Spaces and uses integer linear programming to find the needed similarity values. The work of this paper is focused primarily on space event characterization. In particular, the focus is on the ranking of threats for malicious space events such as a kinetic kill. To make the Conceptual Spaces framework work, the similarity values between the contents of observations on the one hand and the properties of the entities observed on the other needs to be found. This paper shows how to exploit Dempster-Shafer theory to implement a statistical approach for finding these similarities values. This approach will allow a user to identify the uncertainty involved in similarity value data, which can later be propagated through the developed mathematical model in order for the user to know the overall uncertainty in the observation-to-concept mappings needed for space event characterization.

Author Profiles

David Kasmier
University at Buffalo
Barry Smith
University at Buffalo

Analytics

Added to PP
2021-12-31

Downloads
271 (#53,940)

6 months
123 (#24,486)

Historical graph of downloads since first upload
This graph includes both downloads from PhilArchive and clicks on external links on PhilPapers.
How can I increase my downloads?