In the large-scale group decision-making process in food packaging, the social network relationships among decision-making experts have a significant influence on both the decision-making process and outcomes. This study investigates food packaging decision-making method within social network environments and employs probabilistic linguistic term sets to characterize uncertainties in expert evaluations. Secondly, the trust relationship network is constructed based on the trust relationship among decision-making experts, with modularity applied to partition expert communities. Expert weights and index weights are then calculated respectively. Finally, the PROMETHEE II method is applied to the comprehensive sorting of the food packaging alternatives. Through comparative analysis, the feasibility and rationality of the proposed probabilistic linguistic multi-attribute large-scale group decision-making method under social network environment is verified, providing actionable directions for optimizing food packaging solutions.