Multicriterion group classification decision‐making (MCGCDM) involves evaluating alternatives under a specific set of criteria and assigning them to a predetermined set of ordinal categories. One of the main challenges of current research in MCGCDM is establishing a more comprehensive and dynamic feedback adjustment mechanism in the consensus‐reaching process, which ensures the objectivity of the output while better preserving the true intentions of the decision‐makers (DMs). Therefore, this paper constructs a consensus‐driven MCGCDM method based on dynamic trust interaction and overlapping communities’ perspectives. First, the latent factor model algorithm is introduced into the social trust network (STN) based on overlapping community detection to receive the community relations between DMs. Second, according to the trust value and community connection of DMs in STN, a comprehensive, objective, and dynamic experts’ weights calculation method is proposed. Besides, a trust evolution method relying on classification similarity is developed, achieving dynamic updates of trust networks and overlapping communities. A personalized consensus optimization guided by an adaptive adjustment mechanism is designed, emphasizing the significance of overlapping DMs and group opinions while strictly adhering to minimal adjustment. Subsequently, the green building rating is used as an explanatory case to prove the feasibility and objectivity of this method. Eventually, through a series of comparative analyses, the effectiveness and advantages of the proposed method are verified.