Rank-Score Characteristics (RSC) function and cognitive diversity

Abstract

In Combinatorial Fusion Analysis (CFA), a set of multiple scoring systems is used to facilitate integration and fusion of data, features, and/or decisions so as to improve the quality of resultant decisions and actions. Specifically, in a recently developed information fusion method, each system consists of a score function, a rank function, and a Rank-Score Characteristic (RSC) function. The RSC function illustrates the scoring (or ranking) behavior of the system. In this report, we show that RSC functions can be computed easily and RSC functions can be used to measure cognitive diversity for two or more scoring systems. In addition, we show that measuring diversity using the RSC function is inherently distinct from the concept of correlation in statistics and can be used to improve fusion results in classification and decision making. Among a set of domain applications, we discuss information retrieval, virtual screening, and target tracking. © 2010 Springer-Verlag.