Thanks to the expansive development of the Internet based technologies the on-line communities in which millions of users interact in real time is living and apogee. Leverage these networks as tools to carry out massive decision making processes such as the ones involved in e-democracy and e-health communities constitutes not only an extraordinary opportunity but an important research challenge. In this context issues such us influence propagation, agents interaction, and malicious users identification and isolation are key to provide successful solutions to this challenge. In this contribution we aim to address these issues by presenting a new opinion propagation network in which the influence that each agent exert with respect to their neighbours is assessed by means of a combination of the following three aspects: (i)agents’ self-confidence, (ii)inter-agents opinions similarity, (iii) the quality of the information provided by each agent, that is, the lack of contradiction also called as consistency. The proposed network allows to allocate more influence to those agents providing higher quality information and to isolate those who may present a malicious behaviour.