Combining SimTrust and Weighted Simple Exponential Smoothing

TL;DR

This paper proposes combining the SimTrust model, which uses shared interest and opinions for trust, with the Weighted Simple Exponential Smoothing (WSES) trust metric, which relies on explicit ratings, to improve trust assessment in distributed systems.

Contribution

It introduces a novel approach to integrate SimTrust and WSES trust models, analyzing their compatibility and advantages in different scenarios.

Findings

SimTrust is effective with unrated or hard-to-rate items.

WSES performs well with explicit rating data.

The combination enhances trust evaluation flexibility.

Abstract

In the domain of Autonomic and Organic Computing, the entities of a distributed system are variable as well as the efficiency and the intention of their work. Therefore, a scalable mechanism to incentivise/sanction entities which contribute towards/against the system goal is needed. Trust is a suited metric to find benevolent entities. In this paper, we focus for one on the SimTrust model which introduces trust on entities when they share interest and opinions using tagging information. The second model is the Weighted Simple Exponential Smoothing Trust metric (WSES) which functions on explicitly rated items. WSES follows two basic rules which ensure a logic rating mechanism. When putting these two models in context, SimTrust has advantages on items that have not been rated yet or can not easily be rated. WSES is a trust metric which returns good results on explicit rank values. We propose concepts on combining both approaches and state in which cases they are incompatible.