At present, the imperfect network language environment has adversely affected to cultivate the adolescents’ language quality. Therefore, it is urgent to use modern information technology to design a new algorithm model for the cultivation of the adolescents’ language quality. Through in-depth research on the current research status of youth language quality training in recent years, it is found that the introduction of Bayesian network into the cultivation of adolescents’ language quality can effectively improve its training effect. A data mining algorithm and Bayesian network are used to design a BOPPPS model, with the model used to cultivate the language quality of adolescents. To verify whether the BOPPPS model can achieve good results in the cultivation of adolescents’ language quality, a simulation experiment on the BOPPPS model was conducted, evaluating the actual effect by using the multi-level fuzzy comprehensive evaluation model. Results show that the adolescents who are trained with the BOPPS model has significantly improved overall language quality before training, and thus it can be inferred that the BOPPPS model plays a positive role in improving the language quality of adolescents. The conclusions can provide some reference for the cultivation of adolescents’ language quality and has certain theoretical research significance.