The Hegelsmann-Krause Opinion Dynamics model provides a description of the social behaviour of humans among which a decision making process is taking place. In contrast to simpler agreement models, the opinions are not granted to converge to a single value, but may eventually split. Within such a framework, each agent is influenced by the others provided that the difference in their opinion does not exceed a threshold, which is common to all the agents. In this paper, we introduce an extension where each agent is associated with a different threshold value representing the reliability of the information carried by the agent. Moreover, we consider an additional topological constraint in order to take into account also geographical proximity, and the result is a distributed data aggregation and decision making algorithm.