Effects of reputation communication expressiveness in virtual societies

Agents interaction about reputation has to deal with semantic interoperability issues, which can be handled by different approaches using different levels of expressiveness. Previous experiments have already been conducted in order to investigate the effects of a more expressive communication language on agents' reputation evaluation accuracy, but their analyses disregard the possible correlations among reputation models' attributes. Here, we propose the use of a multivariate statistical approach in order to take into account such correlations and to encourage the social simulation community to analyze its experimental outputs using formal mathematical approaches. We also applied the presented approach to the experimental results previously analyzed using a univariate statistical approach. Our analysis corroborate with the latter showing that, in most cases, there is benefit in using a more expressive communication language. Reprinted by permission of Springer