Alternative methods of propagating contradictory evidence

• Authors: Chris J. Hinde;Robert S. Patching;Steve A. McCoy
• Venue: FS'08 Proceedings of the 9th WSEAS International Conference on Fuzzy Systems
• Year: 2008

This paper examines two alternative methods of propagating contradictory evidence in intuitionistic fuzzy sets using mass assignment techniques as the underlying semantics, and proposes a third. Earlier papers assigned contradictory evidence arising from a combinator by considering that contradictory evidence from either operand should result in a contradictory result. The monotonic increase means that a small amount of contradiction results in a very large contradiction component even after a relatively short reasoning path. Adopting an alternative approach that assigns contradictory evidence only where both operands contain contradictory evidence results in monotonically decreasing contradictory components. THis approach would eliminate contradiction in the conclusion giving a specious validity to the result. It is proposed that, ideally a satisfactory propagation method should result in a contradictory measure that is independent of the length of the chain of reasoning. A final algorithm constructed from the earlier two is presented, which is independent of the length of the chain of reasoning. The level of contradiction in the final answer depends only on the contradictory evidence introduced during the reasoning, any contradictions arising directly from the combinations, and not on the length of the chain of reasoning.