For a large group decision-making problem with partly known attribute weights and completely unknown expert weights, a new decision-making method based on the cloud model is proposed. Firstly, an expert weight determination model based on the trust network is established to obtain the weight information of each expert. Secondly, the linguistic preferences with different multi-granularity are transformed into clouds, and then a clustering process is applied. Thirdly, an optimization model is constructed to compute the attribute weights, and then the comprehensive evaluation value for each alternative is presented and the ranking results can be derived. The expert weight determination model can solve the decision making problems where exists a large number of experts and the weight information of each expert is difficult to be provided objectively. In addition, the defined intuitionistic trust set is a good way to describe the trust network among experts, which can help to exploit the information of experts. By transforming the multi-granularity linguistic variables into clouds, it can describe the fuzziness and randomness of linguistic information and avoid information losses and distortions.