Subjective logic with uncertain partial observations

Abstract

Subjective logic (SL) is an effective tool to manage and update beliefs over a set of mutually exclusive assertions. The method to update subjective beliefs from direct observations of assertions is well understood. Recent work has incorporated the SL framework to derive the belief update equations for partial observations where the measurements are only statistically related to the assertions. This work further expands the notion of SL to consider uncertainty in the underlying statistical relationship between measurements and assertions. In other words, new methods are derived for SL that incorporate uncertainty in the reported likelihood of the assertions. Simulations demonstrate the utility of the new likelihood uncertainty aware belief update methods. © 2012 ISIF (Intl Society of Information Fusi).