Applying inferential processes to partner selection in large agents communities

The current literature clearly highlighted the need to define a fast and efficient tool for trust assessment, even in lack of direct information, as much as possessing mechanisms allowing a matching between a selected task and a reliable agent able to carry it out. Direct experience plays a big part, yet it requires a long time to offer a stable and accurate performance and this characteristics may represents a strong drawback especially within huge agents' communities. We support the idea that category-based evaluations and inferential processes represent a useful resource for trust assessment. Within this work, we exploit simulations to investigate how efficient this inferential strategy is, with respect to direct experience, focusing on when and to what extent the first prevails on the latter. Our results suggest that in some situations categories represent a valuable asset, providing even better results.