Multi-view co-attention network for fake news detection by modeling topic-specific user and news source credibility

•We are the first to define topic-specific user credibility based on his socio-cognitive bias and use it for fake news detection.•We present a novel variant of the co-Att mechanism to implicitly model the credibility of users and news sources.•We produce great echo-chamber representation by using BiCM and user news sharing behaviors.•A novel heterogeneous graph is proposed to jointly model news media partisan bias and level of bias with weak supervision.•Our model employs knowledge entities in the news in addition to the partisan bias of its source for more effective representation of news political ideology.